chore(main): release 0.32.0 (#615 )

🤖 I have created a release *beep* *boop* --- ## [0.32.0](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.31.4...v0.32.0) (2026-05-03) ### Features * **deps-dev:** Bump nanoid from 3.3.11 to 3.3.12 ([#622](https://github.com/agentclientprotocol/claude-agent-acp/issues/622)) ([c78ac62](c78ac62d05)) * **deps:** Bump the minor group with 2 updates ([#619](https://github.com/agentclientprotocol/claude-agent-acp/issues/619)) ([6ccd37c](6ccd37ce94)) * **deps:** update to @anthropic-ai/claude-agent-sdk 0.2.126 ([#621](https://github.com/agentclientprotocol/claude-agent-acp/issues/621)) ([becc3b8](becc3b86e6)) * **deps:** Update to @anthropic-ai/claude-agent-sdk@0.2.123 ([#614](https://github.com/agentclientprotocol/claude-agent-acp/issues/614)) ([5b93119](5b9311938d)) ### Bug Fixes * emit a real diff when Write overwrites an existing file ([#618](https://github.com/agentclientprotocol/claude-agent-acp/issues/618)) ([8d7e220](8d7e22026e)) --- This PR was generated with [Release Please](https://github.com/googleapis/release-please). See [documentation](https://github.com/googleapis/release-please#release-please). Co-authored-by: acp-release-bot[bot] <246668977+acp-release-bot[bot]@users.noreply.github.com>
feat(deps): update to @anthropic-ai/claude-agent-sdk 0.2.126 (#621 )
2026-10-27 12:40:32 +00:00 · 2026-05-04 00:02:01 +02:00 · 2026-05-03 23:55:27 +02:00 · 2026-05-03 21:52:06 +00:00 · 2026-05-03 23:44:47 +02:00 · 2026-05-01 19:44:59 +02:00
26 changed files with 7682 additions and 1976 deletions
--- a/.github/dependabot.yml
+++ b/.github/dependabot.yml
@@ -0,0 +1,25 @@
+version: 2
+updates:
+  - package-ecosystem: "github-actions"
+    directory: "/"
+    schedule:
+      # Check for updates to GitHub Actions every week
+      interval: "weekly"
+  - package-ecosystem: "npm"
+    directory: "/"
+    schedule:
+      interval: "daily"
+    allow:
+      - dependency-type: "all"
+    ignore:
+      # Pinned to match the peer range of @anthropic-ai/claude-agent-sdk;
+      # newer minors break type compatibility between the two installed copies.
+      - dependency-name: "@anthropic-ai/sdk"
+        update-types:
+          - "version-update:semver-minor"
+          - "version-update:semver-major"
+    groups:
+      minor:
+        update-types:
+          - "minor"
+          - "patch"
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -21,12 +21,12 @@ jobs:
    timeout-minutes: 5

    steps:
-      - uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd

      - name: Use Node.js
-        uses: actions/setup-node@49933ea5288caeca8642d1e84afbd3f7d6820020
+        uses: actions/setup-node@48b55a011bda9f5d6aeb4c2d9c7362e8dae4041e
        with:
-          node-version: latest
+          node-version: "lts/*"
          cache: "npm"

      - name: Install dependencies
--- a/.github/workflows/publish.yml
+++ b/.github/workflows/publish.yml
@@ -1,92 +1,52 @@
 name: Publish and Release
 on:
  push:
-    tags:
-      - "v*.*.*"
-  workflow_dispatch:
+    branches:
+      - main

 jobs:
-  build-binaries:
-    name: Build (${{ matrix.target }})
+  release-please:
    runs-on: ubuntu-latest
+    environment: release # Optional: for enhanced security
    permissions:
-      contents: read
-    timeout-minutes: 15
-
-    strategy:
-      fail-fast: true
-      matrix:
-        include:
-          - target: bun-linux-x64
-            artifact: claude-agent-acp
-            archive: claude-agent-acp-linux-x64.tar.gz
-          - target: bun-linux-arm64
-            artifact: claude-agent-acp
-            archive: claude-agent-acp-linux-arm64.tar.gz
-          - target: bun-linux-x64-musl
-            artifact: claude-agent-acp
-            archive: claude-agent-acp-linux-x64-musl.tar.gz
-          - target: bun-linux-arm64-musl
-            artifact: claude-agent-acp
-            archive: claude-agent-acp-linux-arm64-musl.tar.gz
-          - target: bun-darwin-x64
-            artifact: claude-agent-acp
-            archive: claude-agent-acp-darwin-x64.zip
-          - target: bun-darwin-arm64
-            artifact: claude-agent-acp
-            archive: claude-agent-acp-darwin-arm64.zip
-          - target: bun-windows-x64
-            artifact: claude-agent-acp.exe
-            archive: claude-agent-acp-windows-x64.zip
-          - target: bun-windows-arm64
-            artifact: claude-agent-acp.exe
-            archive: claude-agent-acp-windows-arm64.zip
-
+      contents: write
+      issues: write
+      pull-requests: write
    steps:
-      - uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8
-
-      - uses: oven-sh/setup-bun@3d267786b128fe76c2f16a390aa2448b815359f3
+      # Generating a GitHub token, so that PRs and tags created by
+      # the release-please-action can trigger actions workflows.
+      - name: Generate GitHub token
+        uses: actions/create-github-app-token@1b10c78c7865c340bc4f6099eb2f838309f1e8c3
+        id: generate-token
        with:
-          bun-version: latest
-
-      - name: Install dependencies
-        run: bun install
-
-      - name: Build binary
-        run: |
-          bun build src/index.ts \
-            --compile \
-            --target ${{ matrix.target }} \
-            --define 'process.env.CLAUDE_AGENT_ACP_IS_SINGLE_FILE_BUN="true"' \
-            --outfile ${{ matrix.artifact }}
-
-      - name: Compress binary
-        run: |
-          if [[ "${{ matrix.archive }}" == *.zip ]]; then
-            zip ${{ matrix.archive }} ${{ matrix.artifact }}
-          else
-            tar czf ${{ matrix.archive }} ${{ matrix.artifact }}
-          fi
-
-      - name: Upload artifact
-        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f
+          # GitHub App ID secret name
+          app-id: ${{ secrets.RELEASE_PLZ_APP_ID }}
+          # GitHub App private key secret name
+          private-key: ${{ secrets.RELEASE_PLZ_APP_PRIVATE_KEY }}
+      - uses: googleapis/release-please-action@45996ed1f6d02564a971a2fa1b5860e934307cf7
+        id: release
        with:
-          name: ${{ matrix.archive }}
-          path: ${{ matrix.archive }}
+          token: ${{ steps.generate-token.outputs.token }}
+          release-type: node
+          config-file: release-please-config.json
+          manifest-file: .release-please-manifest.json
+    outputs:
+      release_created: ${{ steps.release.outputs.release_created }}
+      tag_name: ${{ steps.release.outputs.tag_name }}

  publish-npm:
    name: Publish to npm
    runs-on: ubuntu-latest
-    needs: [build-binaries]
-    if: startsWith(github.ref, 'refs/tags/')
+    needs: [release-please]
+    if: ${{ needs.release-please.outputs.release_created }}
    environment: release # Optional: for enhanced security
    permissions:
      contents: read
      id-token: write
    steps:
-      - uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd
      # Setup .npmrc file to publish to npm
-      - uses: actions/setup-node@49933ea5288caeca8642d1e84afbd3f7d6820020
+      - uses: actions/setup-node@48b55a011bda9f5d6aeb4c2d9c7362e8dae4041e
        with:
          node-version: "lts/*"
          registry-url: "https://registry.npmjs.org"
@@ -95,26 +55,3 @@ jobs:
      - run: npm ci
      - run: npm run build
      - run: npm publish
-
-  release:
-    name: Create Release
-    runs-on: ubuntu-latest
-    needs: [build-binaries, publish-npm]
-    if: startsWith(github.ref, 'refs/tags/')
-    permissions:
-      contents: write
-
-    steps:
-      - uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8
-
-      - name: Download all artifacts
-        uses: actions/download-artifact@37930b1c2abaa49bbe596cd826c3c89aef350131
-        with:
-          path: artifacts
-          merge-multiple: true
-
-      - name: Create Release
-        uses: softprops/action-gh-release@6da8fa9354ddfdc4aeace5fc48d7f679b5214090
-        with:
-          generate_release_notes: true
-          files: artifacts/*
--- a/.prettierignore
+++ b/.prettierignore
@@ -1,3 +1,7 @@
 node_modules
 dist
 .agent-shell
+
+.release-please-manifest.json
+release-please-config.json
+CHANGELOG.md
--- a/.release-please-manifest.json
+++ b/.release-please-manifest.json
@@ -0,0 +1,3 @@
+{
+  ".": "0.32.0"
+}
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,5 +1,237 @@
 # Changelog

+## [0.32.0](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.31.4...v0.32.0) (2026-05-03)
+
+
+### Features
+
+* **deps-dev:** Bump nanoid from 3.3.11 to 3.3.12 ([#622](https://github.com/agentclientprotocol/claude-agent-acp/issues/622)) ([c78ac62](https://github.com/agentclientprotocol/claude-agent-acp/commit/c78ac62d05283e5683b87b198923818f9f556a03))
+* **deps:** Bump the minor group with 2 updates ([#619](https://github.com/agentclientprotocol/claude-agent-acp/issues/619)) ([6ccd37c](https://github.com/agentclientprotocol/claude-agent-acp/commit/6ccd37ce947fbd55691f676344b15ef1266191ad))
+* **deps:** update to @anthropic-ai/claude-agent-sdk 0.2.126 ([#621](https://github.com/agentclientprotocol/claude-agent-acp/issues/621)) ([becc3b8](https://github.com/agentclientprotocol/claude-agent-acp/commit/becc3b86e60eb33e7823a0ef27d5aa99758750d7))
+* **deps:** Update to @anthropic-ai/claude-agent-sdk@0.2.123 ([#614](https://github.com/agentclientprotocol/claude-agent-acp/issues/614)) ([5b93119](https://github.com/agentclientprotocol/claude-agent-acp/commit/5b9311938d97f47400debd66f2b4792e50639c6b))
+
+
+### Bug Fixes
+
+* emit a real diff when Write overwrites an existing file ([#618](https://github.com/agentclientprotocol/claude-agent-acp/issues/618)) ([8d7e220](https://github.com/agentclientprotocol/claude-agent-acp/commit/8d7e22026e15d54208414bd9ad128eb86d41b451))
+
+## [0.31.4](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.31.3...v0.31.4) (2026-04-28)
+
+
+### Bug Fixes
+
+* gate auto mode on model support ([#604](https://github.com/agentclientprotocol/claude-agent-acp/issues/604)) ([ec47d34](https://github.com/agentclientprotocol/claude-agent-acp/commit/ec47d3446fa22e5895621f57c0a4497dbe044505))
+
+## [0.31.3](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.31.2...v0.31.3) (2026-04-28)
+
+
+### Bug Fixes
+
+* Rename effort config category to thought_level ([#609](https://github.com/agentclientprotocol/claude-agent-acp/issues/609)) ([76b4e96](https://github.com/agentclientprotocol/claude-agent-acp/commit/76b4e9650832e59a6b4915ad898fc66a065022fa))
+
+## [0.31.2](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.31.1...v0.31.2) (2026-04-28)
+
+
+### Bug Fixes
+
+* tolerate invalid values in settings.json ([#601](https://github.com/agentclientprotocol/claude-agent-acp/issues/601)) ([67af018](https://github.com/agentclientprotocol/claude-agent-acp/commit/67af018d0216238d8f3cd5bad04a55200518239b))
+
+## [0.31.1](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.31.0...v0.31.1) (2026-04-27)
+
+
+### Bug Fixes
+
+* detect glibc and use correct binary ([#599](https://github.com/agentclientprotocol/claude-agent-acp/issues/599)) ([f0801d5](https://github.com/agentclientprotocol/claude-agent-acp/commit/f0801d50b2390a3c48150b9b8507d30a70c18b84))
+
+## [0.31.0](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.30.0...v0.31.0) (2026-04-24)
+
+
+### Features
+
+* allow configuring models through env vars ([#586](https://github.com/agentclientprotocol/claude-agent-acp/issues/586)) ([1e15f3d](https://github.com/agentclientprotocol/claude-agent-acp/commit/1e15f3d8b0a06076ee62764ab576ac29fe056a38))
+* Support effort levels ([#464](https://github.com/agentclientprotocol/claude-agent-acp/issues/464)) ([9e1185b](https://github.com/agentclientprotocol/claude-agent-acp/commit/9e1185bfd55cbf83a653a0ff8c194f986cbd1574))
+* Update to acp sdk v0.20 ([#589](https://github.com/agentclientprotocol/claude-agent-acp/issues/589)) ([92adcbd](https://github.com/agentclientprotocol/claude-agent-acp/commit/92adcbd1ace2f2103742ffcf5c589eb1ca41b9b8))
+* Update to claude-agent-sdk 0.2.119 ([#587](https://github.com/agentclientprotocol/claude-agent-acp/issues/587)) ([ef1dbd1](https://github.com/agentclientprotocol/claude-agent-acp/commit/ef1dbd1cb457f0524d221ea07235abf49dc941a0))
+
+
+### Bug Fixes
+
+* Forward full systemPrompt preset options from _meta ([#591](https://github.com/agentclientprotocol/claude-agent-acp/issues/591)) ([b3ddfc3](https://github.com/agentclientprotocol/claude-agent-acp/commit/b3ddfc3b86703c23fe99d93611885c12b1e4812c))
+* Session Replay and Permission rendering ([#593](https://github.com/agentclientprotocol/claude-agent-acp/issues/593)) ([5faefab](https://github.com/agentclientprotocol/claude-agent-acp/commit/5faefab8ab9b725f7a3374808bf5579e411884bc)), closes [#579](https://github.com/agentclientprotocol/claude-agent-acp/issues/579)
+
+## [0.30.0](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.29.2...v0.30.0) (2026-04-20)
+
+
+### Features
+
+* Update to claude-agent-sdk 0.2.114 ([#572](https://github.com/agentclientprotocol/claude-agent-acp/issues/572)) ([e9dd452](https://github.com/agentclientprotocol/claude-agent-acp/commit/e9dd45207d396302f20de82561135bf9558c0e76))
+
+## [0.29.2](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.29.1...v0.29.2) (2026-04-17)
+
+
+### Bug Fixes
+
+* Guard against null usage tokens ([#565](https://github.com/agentclientprotocol/claude-agent-acp/issues/565)) ([f7dc300](https://github.com/agentclientprotocol/claude-agent-acp/commit/f7dc300a165e70f6426f53920f12e24c04033cdc)), closes [#564](https://github.com/agentclientprotocol/claude-agent-acp/issues/564)
+
+## [0.29.1](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.29.0...v0.29.1) (2026-04-17)
+
+
+### Bug Fixes
+
+* emit usage updates from Claude stream events ([#506](https://github.com/agentclientprotocol/claude-agent-acp/issues/506)) ([dd67450](https://github.com/agentclientprotocol/claude-agent-acp/commit/dd67450fd9bddc82db3e71273b535e07b2672804))
+* Remove dot from auto mode description ([#561](https://github.com/agentclientprotocol/claude-agent-acp/issues/561)) ([2ecfa83](https://github.com/agentclientprotocol/claude-agent-acp/commit/2ecfa83b26db58deaded210463fa6ff21d0dff70))
+* Update to claude-agent-sdk 0.2.112 to fix Auto bug with Opus 4.7 ([#562](https://github.com/agentclientprotocol/claude-agent-acp/issues/562)) ([079614a](https://github.com/agentclientprotocol/claude-agent-acp/commit/079614ab05afba17e2cce0d3d238df7b90e17389))
+
+## [0.29.0](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.28.0...v0.29.0) (2026-04-16)
+
+
+### Features
+
+* Update to claude-agent-sdk 0.2.111 (Opus 4.7) ([#557](https://github.com/agentclientprotocol/claude-agent-acp/issues/557)) ([85cd70c](https://github.com/agentclientprotocol/claude-agent-acp/commit/85cd70c9f3be47c9c404958547c4046d866db1c9))
+
+## [0.28.0](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.27.0...v0.28.0) (2026-04-15)
+
+
+### Features
+
+* Update to claude-agent-sdk 0.2.109 ([#549](https://github.com/agentclientprotocol/claude-agent-acp/issues/549)) ([07a0fbc](https://github.com/agentclientprotocol/claude-agent-acp/commit/07a0fbc2f6bc388541d064a436412bdd850772cb))
+
+## [0.27.0](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.26.0...v0.27.0) (2026-04-13)
+
+
+### Features
+
+* allow clients to opt into receiving raw SDK messages ([#527](https://github.com/agentclientprotocol/claude-agent-acp/issues/527)) ([403a668](https://github.com/agentclientprotocol/claude-agent-acp/commit/403a668078c067868062ed26cd4d3e36665b66b6))
+* Update to claude-agent-sdk 0.2.104 ([#537](https://github.com/agentclientprotocol/claude-agent-acp/issues/537)) ([6811943](https://github.com/agentclientprotocol/claude-agent-acp/commit/6811943f57ef08616be633a8197223d4072663cf))
+
+
+### Bug Fixes
+
+* Allow auto mode after plan mode and send description for auto mode ([#528](https://github.com/agentclientprotocol/claude-agent-acp/issues/528)) ([fb9aced](https://github.com/agentclientprotocol/claude-agent-acp/commit/fb9aced3151c40694f1f01fd95665c4f5d90eb67))
+* Better remote check for auth methods ([#538](https://github.com/agentclientprotocol/claude-agent-acp/issues/538)) ([93f58c0](https://github.com/agentclientprotocol/claude-agent-acp/commit/93f58c0d2fcf7365c7ca5a6e52f56663e2065ddb))
+* better shutdown logic ([#543](https://github.com/agentclientprotocol/claude-agent-acp/issues/543)) ([9fb631f](https://github.com/agentclientprotocol/claude-agent-acp/commit/9fb631f5d76a5c5f6f0a1b61bdef05fe368c754c))
+* exit process when ACP connection closes ([#530](https://github.com/agentclientprotocol/claude-agent-acp/issues/530)) ([5c81e99](https://github.com/agentclientprotocol/claude-agent-acp/commit/5c81e99fe5ccdf774819b7ff3a2bc78a6519d730))
+* guard tool info rendering when tool_use input is undefined ([#536](https://github.com/agentclientprotocol/claude-agent-acp/issues/536)) ([d627b8c](https://github.com/agentclientprotocol/claude-agent-acp/commit/d627b8c5e95be31ac03923ef67d91748ec8564c5))
+* Remove backup auth check from new session ([#544](https://github.com/agentclientprotocol/claude-agent-acp/issues/544)) ([32b16c1](https://github.com/agentclientprotocol/claude-agent-acp/commit/32b16c1ad99394238b0e5cd6c2f761c12debe142))
+
+## [0.26.0](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.25.3...v0.26.0) (2026-04-08)
+
+
+### Features
+
+* Update claude-agent-sdk to 0.2.96 ([#526](https://github.com/agentclientprotocol/claude-agent-acp/issues/526)) ([c073131](https://github.com/agentclientprotocol/claude-agent-acp/commit/c07313148808a55f27f385c10babbaf7511a7f12))
+
+
+### Bug Fixes
+
+* Remove bun builds from release ([#525](https://github.com/agentclientprotocol/claude-agent-acp/issues/525)) ([fcf5aaf](https://github.com/agentclientprotocol/claude-agent-acp/commit/fcf5aaf06dfe9f7d1b285b976eeb2d1e20ea8dec))
+* Use TUI login for remote environments ([#523](https://github.com/agentclientprotocol/claude-agent-acp/issues/523)) ([cc73e37](https://github.com/agentclientprotocol/claude-agent-acp/commit/cc73e37b41678aa67813c4fbef66ba33ca538743))
+
+## [0.25.3](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.25.2...v0.25.3) (2026-04-06)
+
+
+### Bug Fixes
+
+* Drop claude-agent-sdk back to 0.2.91 to fix broken import ([#513](https://github.com/agentclientprotocol/claude-agent-acp/issues/513)) ([26f3e8a](https://github.com/agentclientprotocol/claude-agent-acp/commit/26f3e8a5216295985fadb80fb3b977045c0c1b2c))
+* Recreate resumed sessions when params change ([#515](https://github.com/agentclientprotocol/claude-agent-acp/issues/515)) ([aa82193](https://github.com/agentclientprotocol/claude-agent-acp/commit/aa82193330026bae132fed8391c47dde777dcf5a))
+
+## [0.25.2](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.25.1...v0.25.2) (2026-04-06)
+
+
+### Bug Fixes
+
+* prioritize ANTHROPIC_MODEL env var over settings.model in model … ([#505](https://github.com/agentclientprotocol/claude-agent-acp/issues/505)) ([bea1a40](https://github.com/agentclientprotocol/claude-agent-acp/commit/bea1a40e9bfe1e06672b118a727f9339def3be23))
+
+## [0.25.1](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.25.0...v0.25.1) (2026-04-06)
+
+
+### Bug Fixes
+
+* add `auto` to valid modes in applySessionMode to fix mode cycling ([#507](https://github.com/agentclientprotocol/claude-agent-acp/issues/507)) ([15e91fb](https://github.com/agentclientprotocol/claude-agent-acp/commit/15e91fb5c3449de2600b583cb7a8d36b5b510443))
+
+## [0.25.0](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.24.2...v0.25.0) (2026-04-03)
+
+
+### Features
+
+* Add auto permission mode support ([#501](https://github.com/agentclientprotocol/claude-agent-acp/issues/501)) ([a161453](https://github.com/agentclientprotocol/claude-agent-acp/commit/a16145396fe2e6ead8734478961b2de65707e335))
+* Add separate Claude and Console terminal logins ([#502](https://github.com/agentclientprotocol/claude-agent-acp/issues/502)) ([063cd35](https://github.com/agentclientprotocol/claude-agent-acp/commit/063cd353809df8f3dda9e57d8ea848c0a6d44257))
+* Update to claude-agent-sdk 0.2.91 ([#500](https://github.com/agentclientprotocol/claude-agent-acp/issues/500)) ([65a2230](https://github.com/agentclientprotocol/claude-agent-acp/commit/65a223038576d72b74e1483fed10e982a1f842bd))
+
+
+### Bug Fixes
+
+* log warnings for malformed settings files instead of silent fallback ([#486](https://github.com/agentclientprotocol/claude-agent-acp/issues/486)) ([ae6c388](https://github.com/agentclientprotocol/claude-agent-acp/commit/ae6c38831415f9fc1de2d3dd1d4a247becbbd32f))
+* prevent race conditions in SettingsManager setCwd and debounce ([#485](https://github.com/agentclientprotocol/claude-agent-acp/issues/485)) ([7506223](https://github.com/agentclientprotocol/claude-agent-acp/commit/7506223cffb1aba4b4560feda11f69a1395a8c9d))
+* use current model's context window for usage_update size ([#412](https://github.com/agentclientprotocol/claude-agent-acp/issues/412)) ([d07799d](https://github.com/agentclientprotocol/claude-agent-acp/commit/d07799d7b3b4e438c8b158266c79723a0b592c07))
+
+## [0.24.2](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.24.1...v0.24.2) (2026-03-27)
+
+
+### Bug Fixes
+
+* Add explicit type checks for MCP servers (http/sse) ([#487](https://github.com/agentclientprotocol/claude-agent-acp/issues/487)) ([e00a439](https://github.com/agentclientprotocol/claude-agent-acp/commit/e00a43901fa2b4fd7d582e1de57277be88233007))
+
+## [0.24.1](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.24.0...v0.24.1) (2026-03-26)
+
+
+### Bug Fixes
+
+* Cleanup based on new idle state [#463](https://github.com/agentclientprotocol/claude-agent-acp/issues/463) ([#480](https://github.com/agentclientprotocol/claude-agent-acp/issues/480)) ([23b3073](https://github.com/agentclientprotocol/claude-agent-acp/commit/23b30730253752f0bc4e30b619a6236f16fafdb9))
+
+## 0.24.0
+
+Rename from `@zed-industries/claude-agent-acp` to `@agentclientprotocol/claude-agent-acp`.
+
+We are moving this to the main ACP org to better allow multiple teams to contribute and maintain this adapter.
+
+## 0.23.1
+
+- Add back error_during_execution break point (#469)
+
+## 0.23.0
+
+- Use idle session state as end of turn (#463)
+- Update claude-agent-sdk to 0.2.83 (#462)
+- Fix handling of local-only slash commands (#432)
+- fix: correct null check for gatewayAuthMeta in subscription validation (#455)
+- fix: include both stdout and stderr in Bash tool output (#456)
+- fix: prevent prompt loop hang when cancel races with first result (#458)
+- fix: dispose SettingsManager on session close to prevent resource leaks (#454)
+- fix: restore plan content in ExitPlanMode tool call (#451)
+
+## 0.22.2
+
+- Add experimental meta param for testing additional directories
+
+## 0.22.1
+
+- Fix: invalid auth required state in gateway mode
+
+## 0.22.0
+
+- Use stable list sessions method (#429)
+- Use correct Claude CLI path for static binaries (#428)
+- Update claude-agent-sdk to 0.2.76 (#427)
+- fix: resolve model aliases in setSessionConfigOption (#401) (#403)
+- Reuse existing sessions for load/resume if possible (#426)
+- Remove interrupt flag from deny responses (#425)
+- Allow Bypass permissions mode after Exiting plan (#410)
+- Don't get out of sync when background task creates new init/result (try 2) (#400)
+- Add session/close support (#409)
+
+## 0.21.0
+
+- Update to claude-agent-sdk 0.2.71
+- show project-relative paths in tool call titles
+- Lib: pass through tools array to control built-in tool availability
+- fix: skip user replay
+- fix: handle renamed Agent tool in toolInfoFromToolUse
+
+## 0.20.2
+
+- Update to @anthropic-ai/claude-agent-sdk@0.2.68
+
 ## 0.20.1

 - fix: inherit process.env when spawning agent subprocess
--- a/61
+++ b/61
@@ -1,42 +1,18 @@
-Copyright 2022 - 2025 Zed Industries, Inc.

-
-   Licensed under the Apache License, Version 2.0 (the "License");
-   you may not use this file except in compliance with the License.
-   You may obtain a copy of the License at
-
-
-       http://www.apache.org/licenses/LICENSE-2.0
-
-
-   Unless required by applicable law or agreed to in writing, software
-   distributed under the License is distributed on an "AS IS" BASIS,
-   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-   See the License for the specific language governing permissions and
-   limitations under the License.
-
-
-
-
-Apache License
+                                 Apache License
                           Version 2.0, January 2004
                        http://www.apache.org/licenses/

-
   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION

-
   1. Definitions.

-
      "License" shall mean the terms and conditions for use, reproduction,
      and distribution as defined by Sections 1 through 9 of this document.

-
      "Licensor" shall mean the copyright owner or entity authorized by
      the copyright owner that is granting the License.

-
      "Legal Entity" shall mean the union of the acting entity and all
      other entities that control, are controlled by, or are under common
      control with that entity. For the purposes of this definition,
@@ -45,28 +21,23 @@ Apache License
      otherwise, or (ii) ownership of fifty percent (50%) or more of the
      outstanding shares, or (iii) beneficial ownership of such entity.

-
      "You" (or "Your") shall mean an individual or Legal Entity
      exercising permissions granted by this License.

-
      "Source" form shall mean the preferred form for making modifications,
      including but not limited to software source code, documentation
      source, and configuration files.

-
      "Object" form shall mean any form resulting from mechanical
      transformation or translation of a Source form, including but
      not limited to compiled object code, generated documentation,
      and conversions to other media types.

-
      "Work" shall mean the work of authorship, whether in Source or
      Object form, made available under the License, as indicated by a
      copyright notice that is included in or attached to the work
      (an example is provided in the Appendix below).

-
      "Derivative Works" shall mean any work, whether in Source or Object
      form, that is based on (or derived from) the Work and for which the
      editorial revisions, annotations, elaborations, or other modifications
@@ -75,7 +46,6 @@ Apache License
      separable from, or merely link (or bind by name) to the interfaces of,
      the Work and Derivative Works thereof.

-
      "Contribution" shall mean any work of authorship, including
      the original version of the Work and any modifications or additions
      to that Work or Derivative Works thereof, that is intentionally
@@ -90,12 +60,10 @@ Apache License
      excluding communication that is conspicuously marked or otherwise
      designated in writing by the copyright owner as "Not a Contribution."

-
      "Contributor" shall mean Licensor and any individual or Legal Entity
      on behalf of whom a Contribution has been received by Licensor and
      subsequently incorporated within the Work.

-
   2. Grant of Copyright License. Subject to the terms and conditions of
      this License, each Contributor hereby grants to You a perpetual,
      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
@@ -103,7 +71,6 @@ Apache License
      publicly display, publicly perform, sublicense, and distribute the
      Work and such Derivative Works in Source or Object form.

-
   3. Grant of Patent License. Subject to the terms and conditions of
      this License, each Contributor hereby grants to You a perpetual,
      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
@@ -120,28 +87,23 @@ Apache License
      granted to You under this License for that Work shall terminate
      as of the date such litigation is filed.

-
   4. Redistribution. You may reproduce and distribute copies of the
      Work or Derivative Works thereof in any medium, with or without
      modifications, and in Source or Object form, provided that You
      meet the following conditions:

-
      (a) You must give any other recipients of the Work or
          Derivative Works a copy of this License; and

-
      (b) You must cause any modified files to carry prominent notices
          stating that You changed the files; and

-
      (c) You must retain, in the Source form of any Derivative Works
          that You distribute, all copyright, patent, trademark, and
          attribution notices from the Source form of the Work,
          excluding those notices that do not pertain to any part of
          the Derivative Works; and

-
      (d) If the Work includes a "NOTICE" text file as part of its
          distribution, then any Derivative Works that You distribute must
          include a readable copy of the attribution notices contained
@@ -159,7 +121,6 @@ Apache License
          that such additional attribution notices cannot be construed
          as modifying the License.

-
      You may add Your own copyright statement to Your modifications and
      may provide additional or different license terms and conditions
      for use, reproduction, or distribution of Your modifications, or
@@ -167,7 +128,6 @@ Apache License
      reproduction, and distribution of the Work otherwise complies with
      the conditions stated in this License.

-
   5. Submission of Contributions. Unless You explicitly state otherwise,
      any Contribution intentionally submitted for inclusion in the Work
      by You to the Licensor shall be under the terms and conditions of
@@ -176,13 +136,11 @@ Apache License
      the terms of any separate license agreement you may have executed
      with Licensor regarding such Contributions.

-
   6. Trademarks. This License does not grant permission to use the trade
      names, trademarks, service marks, or product names of the Licensor,
      except as required for reasonable and customary use in describing the
      origin of the Work and reproducing the content of the NOTICE file.

-
   7. Disclaimer of Warranty. Unless required by applicable law or
      agreed to in writing, Licensor provides the Work (and each
      Contributor provides its Contributions) on an "AS IS" BASIS,
@@ -193,7 +151,6 @@ Apache License
      appropriateness of using or redistributing the Work and assume any
      risks associated with Your exercise of permissions under this License.

-
   8. Limitation of Liability. In no event and under no legal theory,
      whether in tort (including negligence), contract, or otherwise,
      unless required by applicable law (such as deliberate and grossly
@@ -206,7 +163,6 @@ Apache License
      other commercial damages or losses), even if such Contributor
      has been advised of the possibility of such damages.

-
   9. Accepting Warranty or Additional Liability. While redistributing
      the Work or Derivative Works thereof, You may choose to offer,
      and charge a fee for, acceptance of support, warranty, indemnity,
@@ -218,5 +174,18 @@ Apache License
      incurred by, or claims asserted against, such Contributor by reason
      of your accepting any such warranty or additional liability.

-
   END OF TERMS AND CONDITIONS
+
+   Copyright 2025 Zed Industries, Inc. and contributors
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.
--- a/README.md
+++ b/README.md
@@ -1,8 +1,8 @@
 # ACP adapter for the Claude Agent SDK

-[![npm](https://img.shields.io/npm/v/%40zed-industries%2Fclaude-agent-acp)](https://www.npmjs.com/package/@zed-industries/claude-agent-acp)
+[![npm](https://img.shields.io/npm/v/%40agentclientprotocol%2Fclaude-agent-acp)](https://www.npmjs.com/package/@agentclientprotocol/claude-agent-acp)

-Use [Claude Agent SDK](https://platform.claude.com/docs/en/agent-sdk/overview#branding-guidelines) from [ACP-compatible](https://agentclientprotocol.com) clients such as [Zed](https://zed.dev)!
+Use [Claude Agent SDK](https://platform.claude.com/docs/en/agent-sdk/overview#branding-guidelines) from [ACP-compatible](https://agentclientprotocol.com) clients!

 This tool implements an ACP agent by using the official [Claude Agent SDK](https://platform.claude.com/docs/en/agent-sdk/overview), supporting:

@@ -18,41 +18,8 @@ This tool implements an ACP agent by using the official [Claude Agent SDK](https

 Learn more about the [Agent Client Protocol](https://agentclientprotocol.com/).

-## How to use
+## Contribution Policy

-### Zed
+This project does not require a Contributor License Agreement (CLA). Instead, contributions are accepted under the following terms:

-The latest version of Zed can already use this adapter out of the box.
-
-To use Claude Agent, open the Agent Panel and click "New Claude Agent Thread" from the `+` button menu in the top-right:
-
-https://github.com/user-attachments/assets/ddce66c7-79ac-47a3-ad59-4a6a3ca74903
-
-Read the docs on [External Agent](https://zed.dev/docs/ai/external-agents) support.
-
-### Other clients
-
-Or try it with any of the other [ACP compatible clients](https://agentclientprotocol.com/overview/clients)!
-
-#### Installation
-
-Install the adapter from `npm`:
-
-```bash
-npm install -g @zed-industries/claude-agent-acp
-```
-
-You can then use `claude-agent-acp` as a regular ACP agent:
-
-```
-ANTHROPIC_API_KEY=sk-... claude-agent-acp
-```
-
-#### Single-file executable
-
-Pre-built single-file binaries are available on the [Releases](https://github.com/zed-industries/claude-agent-acp/releases) page for Linux, macOS, and Windows.
-These binaries bundle everything needed and don't require Node.js.
-
-## License
-
-Apache-2.0
+> By contributing to this project, you agree that your contributions will be licensed under the [Apache License, Version 2.0](https://www.apache.org/licenses/LICENSE-2.0). You affirm that you have the legal right to submit your work, that you are not including code you do not have rights to, and that you understand contributions are made without requiring a Contributor License Agreement (CLA).
--- a/docs/model-configuration.md
+++ b/docs/model-configuration.md
@@ -0,0 +1,56 @@
+# Model Configuration
+
+When using claude-agent-acp with alternative providers (e.g. AWS Bedrock), model IDs differ from the direct Anthropic API. The `CLAUDE_MODEL_CONFIG` environment variable lets you configure model overrides and availability at the deployment level.
+
+## `CLAUDE_MODEL_CONFIG`
+
+A JSON string with two optional fields:
+
+| Field             | Type                     | Description                                                                                                   |
+| ----------------- | ------------------------ | ------------------------------------------------------------------------------------------------------------- |
+| `modelOverrides`  | `Record<string, string>` | Maps Anthropic model IDs to provider-specific model IDs (e.g. Bedrock model IDs or ARNs)                      |
+| `availableModels` | `string[]`               | Restricts which models are offered to users. Accepts aliases (`"opus"`), prefixes (`"opus-4-5"`), or full IDs |
+
+### Examples
+
+**Bedrock model overrides:**
+
+```bash
+CLAUDE_MODEL_CONFIG='{"modelOverrides":{"claude-opus-4-6":"us.anthropic.claude-opus-4-6-v1","claude-sonnet-4-5":"us.anthropic.claude-sonnet-4-5-v1"}}'
+```
+
+**Restrict available models:**
+
+```bash
+CLAUDE_MODEL_CONFIG='{"availableModels":["opus","sonnet"]}'
+```
+
+**Both together:**
+
+```bash
+CLAUDE_MODEL_CONFIG='{"modelOverrides":{"claude-opus-4-6":"us.anthropic.claude-opus-4-6-v1","claude-sonnet-4-5":"us.anthropic.claude-sonnet-4-5-v1"},"availableModels":["opus","sonnet"]}'
+```
+
+**Full Bedrock example:**
+
+```bash
+CLAUDE_CODE_USE_BEDROCK=1 \
+AWS_REGION=us-west-2 \
+CLAUDE_MODEL_CONFIG='{"modelOverrides":{"claude-opus-4-6":"us.anthropic.claude-opus-4-6-v1"}}' \
+node dist/index.js
+```
+
+## Precedence
+
+When an ACP caller provides `settings` via `_meta.claudeCode.options.settings` in the `sessions/create` request, `CLAUDE_MODEL_CONFIG` is ignored entirely. The env var is a deployment-level fallback for cases where the caller does not configure model settings itself.
+
+| Source                                       | Priority                                              |
+| -------------------------------------------- | ----------------------------------------------------- |
+| `_meta.claudeCode.options.settings` (caller) | Highest — used if present                             |
+| `CLAUDE_MODEL_CONFIG` (env var)              | Fallback — used only when caller provides no settings |
+
+## Format details
+
+- The value must be valid JSON. Invalid JSON will cause session creation to fail with a parse error.
+- Only `modelOverrides` and `availableModels` keys are read; other keys in the JSON are ignored.
+- Both fields map directly to the Claude Agent SDK's `Settings` type.
--- a/package-lock.json
+++ b/package-lock.json
--- a/package.json
+++ b/package.json
@@ -1,9 +1,9 @@
 {
-  "name": "@zed-industries/claude-agent-acp",
+  "name": "@agentclientprotocol/claude-agent-acp",
  "publishConfig": {
    "access": "public"
  },
-  "version": "0.20.1",
+  "version": "0.32.0",
  "description": "An ACP-compatible coding agent powered by the Claude Agent SDK (TypeScript)",
  "main": "dist/lib.js",
  "types": "dist/lib.d.ts",
@@ -49,33 +49,33 @@
    "sdk",
    "code"
  ],
-  "homepage": "https://github.com/zed-industries/claude-agent-acp#readme",
+  "homepage": "https://github.com/agentclientprotocol/claude-agent-acp#readme",
  "bugs": {
-    "url": "https://github.com/zed-industries/claude-agent-acp/issues"
+    "url": "https://github.com/agentclientprotocol/claude-agent-acp/issues"
  },
  "repository": {
    "type": "git",
-    "url": "git+https://github.com/zed-industries/claude-agent-acp.git"
+    "url": "git+https://github.com/agentclientprotocol/claude-agent-acp.git"
  },
  "author": "Zed Industries",
  "license": "Apache-2.0",
  "dependencies": {
-    "@agentclientprotocol/sdk": "0.14.1",
-    "@anthropic-ai/claude-agent-sdk": "0.2.63",
+    "@agentclientprotocol/sdk": "0.21.0",
+    "@anthropic-ai/claude-agent-sdk": "0.2.126",
    "zod": "^3.25.0 || ^4.0.0"
  },
  "devDependencies": {
-    "@anthropic-ai/sdk": "0.78.0",
+    "@anthropic-ai/sdk": "0.89.0",
    "@eslint/js": "10.0.1",
-    "@types/node": "25.3.3",
-    "@typescript-eslint/eslint-plugin": "8.56.1",
-    "@typescript-eslint/parser": "8.56.1",
-    "eslint": "10.0.2",
+    "@types/node": "25.6.0",
+    "@typescript-eslint/eslint-plugin": "8.59.1",
+    "@typescript-eslint/parser": "8.59.1",
+    "eslint": "10.3.0",
    "eslint-config-prettier": "10.1.8",
-    "globals": "17.4.0",
-    "prettier": "3.8.1",
+    "globals": "17.6.0",
+    "prettier": "3.8.3",
    "ts-node": "10.9.2",
-    "typescript": "5.9.3",
-    "vitest": "4.0.18"
+    "typescript": "6.0.3",
+    "vitest": "4.1.5"
  }
 }
--- a/release-please-config.json
+++ b/release-please-config.json
@@ -0,0 +1,13 @@
+{
+  "bump-minor-pre-major": true,
+  "bump-patch-for-minor-pre-major": true,
+  "packages": {
+    ".": {
+      "changelog-path": "CHANGELOG.md",
+      "release-type": "node",
+      "bump-minor-pre-major": true,
+      "bump-patch-for-minor-pre-major": true
+    }
+  },
+  "$schema": "https://raw.githubusercontent.com/googleapis/release-please/main/schemas/config.json"
+}
--- a/src/acp-agent.ts
+++ b/src/acp-agent.ts
--- a/src/index.ts
+++ b/src/index.ts
@@ -2,12 +2,38 @@

 // Load managed settings and apply environment variables
 import { loadManagedSettings, applyEnvironmentSettings } from "./utils.js";
-import { runAcp } from "./acp-agent.js";
+import { claudeCliPath, runAcp } from "./acp-agent.js";

 if (process.argv.includes("--cli")) {
-  process.argv = process.argv.filter((arg) => arg !== "--cli");
-  // @ts-expect-error -- no types
-  await import("@anthropic-ai/claude-agent-sdk/cli.js");
+  const { spawn } = await import("node:child_process");
+  const args = process.argv.slice(2).filter((arg) => arg !== "--cli");
+  const child = spawn(await claudeCliPath(), args, { stdio: "inherit" });
+
+  const signals =
+    process.platform === "win32"
+      ? (["SIGINT", "SIGTERM"] as const)
+      : (["SIGINT", "SIGTERM", "SIGHUP"] as const);
+  for (const sig of signals) {
+    process.on(sig, () => {
+      if (!child.killed) child.kill(sig);
+    });
+  }
+
+  child.on("exit", (code, signal) => {
+    if (signal && process.platform !== "win32") {
+      // Remove our listener so re-raising actually terminates instead of
+      // re-entering the no-op handler, which would let us exit with code 0
+      // instead of the signal's conventional 128+N.
+      process.removeAllListeners(signal);
+      process.kill(process.pid, signal);
+    } else {
+      process.exit(code ?? 1);
+    }
+  });
+  child.on("error", (err) => {
+    console.error(err);
+    process.exit(1);
+  });
 } else {
  const managedSettings = loadManagedSettings();
  if (managedSettings) {
@@ -25,8 +51,23 @@ if (process.argv.includes("--cli")) {
    console.error("Unhandled Rejection at:", promise, "reason:", reason);
  });

-  runAcp();
+  const { connection, agent } = runAcp();

-  // Keep process alive
+  async function shutdown() {
+    await agent.dispose().catch((err) => {
+      console.error("Error during cleanup:", err);
+    });
+    process.exit(0);
+  }
+
+  // Exit cleanly when the ACP connection closes (e.g. stdin EOF, transport
+  // error). Without this, `process.stdin.resume()` keeps the event loop
+  // alive indefinitely, causing orphan process accumulation in oneshot mode.
+  connection.closed.then(shutdown);
+
+  process.on("SIGTERM", shutdown);
+  process.on("SIGINT", shutdown);
+
+  // Keep process alive while connection is open
  process.stdin.resume();
 }
--- a/src/lib.ts
+++ b/src/lib.ts
@@ -1,11 +1,14 @@
 // Export the main agent class and utilities for library usage
 export {
  ClaudeAcpAgent,
+  isLocalCommandMetadata,
+  stripLocalCommandMetadata,
  runAcp,
  toAcpNotifications,
  streamEventToAcpNotifications,
  type ToolUpdateMeta,
  type NewSessionMeta,
+  type SDKMessageFilter,
 } from "./acp-agent.js";
 export {
  loadManagedSettings,
@@ -15,7 +18,12 @@ export {
  Pushable,
  unreachable,
 } from "./utils.js";
-export { toolInfoFromToolUse, planEntries, toolUpdateFromToolResult } from "./tools.js";
+export {
+  toolInfoFromToolUse,
+  toDisplayPath,
+  planEntries,
+  toolUpdateFromToolResult,
+} from "./tools.js";
 export {
  SettingsManager,
  type ClaudeCodeSettings,
--- a/src/settings.ts
+++ b/src/settings.ts
@@ -22,12 +22,18 @@ export interface ClaudeCodeSettings {
  permissions?: PermissionSettings;
  env?: Record<string, string>;
  model?: string;
+  effortLevel?: string;
 }

 /**
- * Reads and parses a JSON settings file, returning an empty object if not found or invalid
+ * Reads and parses a JSON settings file, returning an empty object if not found or invalid.
+ * Silently ignores missing files (ENOENT) but logs warnings for other errors
+ * (malformed JSON, permission errors, etc.) to aid debugging.
 */
-async function loadSettingsFile(filePath: string | null): Promise<ClaudeCodeSettings> {
+async function loadSettingsFile(
+  filePath: string | null,
+  logger?: { error: (...args: any[]) => void },
+): Promise<ClaudeCodeSettings> {
  if (!filePath) {
    return {};
  }
@@ -35,7 +41,11 @@ async function loadSettingsFile(filePath: string | null): Promise<ClaudeCodeSett
  try {
    const content = await fs.promises.readFile(filePath, "utf-8");
    return JSON.parse(content) as ClaudeCodeSettings;
-  } catch {
+  } catch (error) {
+    if (error instanceof Error && "code" in error && error.code === "ENOENT") {
+      return {};
+    }
+    logger?.error(`Failed to load settings from ${filePath}:`, error);
    return {};
  }
 }
@@ -83,7 +93,9 @@ export class SettingsManager {
  private onChange?: () => void;
  private logger: { log: (...args: any[]) => void; error: (...args: any[]) => void };
  private initialized = false;
+  private disposed = false;
  private debounceTimer: ReturnType<typeof setTimeout> | null = null;
+  private initPromise: Promise<void> | null = null;

  constructor(cwd: string, options?: SettingsManagerOptions) {
    this.cwd = cwd;
@@ -98,10 +110,19 @@ export class SettingsManager {
    if (this.initialized) {
      return;
    }
+    if (this.initPromise) {
+      return this.initPromise;
+    }

-    await this.loadAllSettings();
-    this.setupWatchers();
-    this.initialized = true;
+    this.disposed = false;
+    this.initPromise = this.loadAllSettings().then(() => {
+      if (!this.disposed) {
+        this.setupWatchers();
+        this.initialized = true;
+      }
+      this.initPromise = null;
+    });
+    return this.initPromise;
  }

  /**
@@ -130,10 +151,10 @@ export class SettingsManager {
   */
  private async loadAllSettings(): Promise<void> {
    const [userSettings, projectSettings, localSettings, enterpriseSettings] = await Promise.all([
-      loadSettingsFile(this.getUserSettingsPath()),
-      loadSettingsFile(this.getProjectSettingsPath()),
-      loadSettingsFile(this.getLocalSettingsPath()),
-      loadSettingsFile(getManagedSettingsPath()),
+      loadSettingsFile(this.getUserSettingsPath(), this.logger),
+      loadSettingsFile(this.getProjectSettingsPath(), this.logger),
+      loadSettingsFile(this.getLocalSettingsPath(), this.logger),
+      loadSettingsFile(getManagedSettingsPath(), this.logger),
    ]);

    this.userSettings = userSettings;
@@ -168,6 +189,10 @@ export class SettingsManager {
        merged.model = settings.model;
      }

+      if (settings.effortLevel !== undefined) {
+        merged.effortLevel = settings.effortLevel;
+      }
+
      if (settings.permissions?.defaultMode !== undefined) {
        merged.permissions = {
          ...merged.permissions,
@@ -226,9 +251,14 @@ export class SettingsManager {

    this.debounceTimer = setTimeout(async () => {
      this.debounceTimer = null;
+      if (this.disposed) {
+        return;
+      }
      try {
        await this.loadAllSettings();
-        this.onChange?.();
+        if (!this.disposed) {
+          this.onChange?.();
+        }
      } catch (error) {
        this.logger.error("Failed to reload settings:", error);
      }
@@ -259,7 +289,6 @@ export class SettingsManager {

    this.dispose();
    this.cwd = cwd;
-    this.initialized = false;
    await this.initialize();
  }

@@ -267,6 +296,10 @@ export class SettingsManager {
   * Disposes of file watchers and cleans up resources
   */
  dispose(): void {
+    this.disposed = true;
+    this.initialized = false;
+    this.initPromise = null;
+
    if (this.debounceTimer) {
      clearTimeout(this.debounceTimer);
      this.debounceTimer = null;
@@ -276,6 +309,5 @@ export class SettingsManager {
      watcher.close();
    }
    this.watchers = [];
-    this.initialized = false;
  }
 }
--- a/src/tests/acp-agent-settings.test.ts
+++ b/src/tests/acp-agent-settings.test.ts
@@ -39,7 +39,7 @@ describe("ClaudeAcpAgent settings", () => {
        initializationResult: async () => ({
          models: [
            {
-              value: "claude-sonnet-4-5",
+              value: "claude-sonnet-4-6",
              displayName: "Claude Sonnet 4.5",
              description: "Default",
            },
@@ -145,12 +145,12 @@ describe("ClaudeAcpAgent settings", () => {
    expect(response.modes.currentModeId).toBe("default");
  });

-  it("throws when permissions.defaultMode is not a string", async () => {
+  it("falls back to 'default' when permissions.defaultMode is invalid", async () => {
    await fs.promises.writeFile(
      path.join(tempDir, "settings.json"),
      JSON.stringify({
        permissions: {
-          defaultMode: 123,
+          defaultMode: "not-a-real-mode",
        },
      }),
    );
@@ -158,18 +158,208 @@ describe("ClaudeAcpAgent settings", () => {
    const projectDir = path.join(tempDir, "project");
    await fs.promises.mkdir(projectDir, { recursive: true });

-    mockQuery();
+    const { getCapturedOptions } = mockQuery();

    const { ClaudeAcpAgent } = await import("../acp-agent.js");
    const agent: ClaudeAcpAgentType = new ClaudeAcpAgent(createMockClient());

-    await expect(
-      (agent as any).createSession({
+    const response = await (agent as any).createSession({
+      cwd: projectDir,
+      mcpServers: [],
+      _meta: { disableBuiltInTools: true },
+    });
+
+    // Bad mode is ignored at the usage site; session creation must not throw.
+    expect(getCapturedOptions().permissionMode).toBe("default");
+    expect(response.modes.currentModeId).toBe("default");
+  });
+
+  it("ignores model from settings when it is not a string", async () => {
+    await fs.promises.writeFile(
+      path.join(tempDir, "settings.json"),
+      JSON.stringify({
+        model: 123,
+      }),
+    );
+
+    const projectDir = path.join(tempDir, "project");
+    await fs.promises.mkdir(projectDir, { recursive: true });
+
+    const { setModelSpy } = mockQuery();
+
+    const { ClaudeAcpAgent } = await import("../acp-agent.js");
+    const agent: ClaudeAcpAgentType = new ClaudeAcpAgent(createMockClient());
+
+    const response = await (agent as any).createSession({
+      cwd: projectDir,
+      mcpServers: [],
+      _meta: { disableBuiltInTools: true },
+    });
+
+    // Bad model is ignored at the usage site; falls back to the first SDK model.
+    expect(setModelSpy).toHaveBeenCalledWith("claude-sonnet-4-6");
+    expect(response.models.currentModelId).toBe("claude-sonnet-4-6");
+  });
+
+  describe("auto mode availability per model", () => {
+    function mockQueryWithModels(models: any[]): {
+      getCapturedOptions: () => any;
+      setModelSpy: ReturnType<typeof vi.fn>;
+      setPermissionModeSpy: ReturnType<typeof vi.fn>;
+    } {
+      let capturedOptions: any;
+      const setModelSpy = vi.fn();
+      const setPermissionModeSpy = vi.fn();
+      querySpy.mockImplementation(({ options }: any) => {
+        capturedOptions = options;
+        return {
+          initializationResult: async () => ({ models }),
+          setModel: setModelSpy,
+          setPermissionMode: setPermissionModeSpy,
+          supportedCommands: async () => [],
+        } as any;
+      });
+      return {
+        getCapturedOptions: () => capturedOptions,
+        setModelSpy,
+        setPermissionModeSpy,
+      };
+    }
+
+    it("omits `auto` from availableModes when the resolved model lacks supportsAutoMode", async () => {
+      const projectDir = path.join(tempDir, "project");
+      await fs.promises.mkdir(projectDir, { recursive: true });
+
+      mockQueryWithModels([
+        {
+          value: "claude-haiku-4-5",
+          displayName: "Claude Haiku",
+          description: "Fast",
+          // supportsAutoMode intentionally omitted
+        },
+      ]);
+
+      const { ClaudeAcpAgent } = await import("../acp-agent.js");
+      const agent: ClaudeAcpAgentType = new ClaudeAcpAgent(createMockClient());
+
+      const response = await (agent as any).createSession({
        cwd: projectDir,
        mcpServers: [],
        _meta: { disableBuiltInTools: true },
-      }),
-    ).rejects.toThrow("Invalid permissions.defaultMode");
+      });
+
+      const modeIds: string[] = response.modes.availableModes.map((m: any) => m.id);
+      expect(modeIds).not.toContain("auto");
+      expect(modeIds).toEqual(
+        expect.arrayContaining(["default", "acceptEdits", "plan", "dontAsk"]),
+      );
+    });
+
+    it("includes `auto` when the resolved model has supportsAutoMode: true", async () => {
+      const projectDir = path.join(tempDir, "project");
+      await fs.promises.mkdir(projectDir, { recursive: true });
+
+      mockQueryWithModels([
+        {
+          value: "claude-opus-4-5",
+          displayName: "Claude Opus",
+          description: "Most capable",
+          supportsAutoMode: true,
+        },
+      ]);
+
+      const { ClaudeAcpAgent } = await import("../acp-agent.js");
+      const agent: ClaudeAcpAgentType = new ClaudeAcpAgent(createMockClient());
+
+      const response = await (agent as any).createSession({
+        cwd: projectDir,
+        mcpServers: [],
+        _meta: { disableBuiltInTools: true },
+      });
+
+      const modeIds: string[] = response.modes.availableModes.map((m: any) => m.id);
+      expect(modeIds).toContain("auto");
+    });
+
+    it("clamps permissions.defaultMode='auto' to 'default' on a model that lacks supportsAutoMode", async () => {
+      await fs.promises.writeFile(
+        path.join(tempDir, "settings.json"),
+        JSON.stringify({ permissions: { defaultMode: "auto" } }),
+      );
+
+      const projectDir = path.join(tempDir, "project");
+      await fs.promises.mkdir(projectDir, { recursive: true });
+
+      const { getCapturedOptions, setPermissionModeSpy } = mockQueryWithModels([
+        {
+          value: "claude-haiku-4-5",
+          displayName: "Claude Haiku",
+          description: "Fast",
+          // supportsAutoMode intentionally omitted
+        },
+      ]);
+
+      const errorSpy = vi.spyOn(console, "error").mockImplementation(() => {});
+
+      try {
+        const { ClaudeAcpAgent } = await import("../acp-agent.js");
+        const agent: ClaudeAcpAgentType = new ClaudeAcpAgent(createMockClient());
+
+        const response = await (agent as any).createSession({
+          cwd: projectDir,
+          mcpServers: [],
+          _meta: { disableBuiltInTools: true },
+        });
+
+        // Options.permissionMode is built before init resolves, so it still
+        // carries the user-typed value; the SDK was synced via
+        // setPermissionMode after we discovered the model can't honor it.
+        expect(getCapturedOptions().permissionMode).toBe("auto");
+        expect(setPermissionModeSpy).toHaveBeenCalledWith("default");
+        expect(response.modes.currentModeId).toBe("default");
+        expect(response.modes.availableModes.map((m: any) => m.id)).not.toContain("auto");
+
+        // A descriptive warning was logged so operators see the clamp.
+        const messages = errorSpy.mock.calls.map((c) => c.join(" "));
+        expect(messages.some((m) => m.includes("auto") && m.includes("claude-haiku-4-5"))).toBe(
+          true,
+        );
+      } finally {
+        errorSpy.mockRestore();
+      }
+    });
+
+    it("does not clamp permissions.defaultMode='auto' on a model that supports auto", async () => {
+      await fs.promises.writeFile(
+        path.join(tempDir, "settings.json"),
+        JSON.stringify({ permissions: { defaultMode: "auto" } }),
+      );
+
+      const projectDir = path.join(tempDir, "project");
+      await fs.promises.mkdir(projectDir, { recursive: true });
+
+      const { getCapturedOptions, setPermissionModeSpy } = mockQueryWithModels([
+        {
+          value: "claude-opus-4-5",
+          displayName: "Claude Opus",
+          description: "Most capable",
+          supportsAutoMode: true,
+        },
+      ]);
+
+      const { ClaudeAcpAgent } = await import("../acp-agent.js");
+      const agent: ClaudeAcpAgentType = new ClaudeAcpAgent(createMockClient());
+
+      const response = await (agent as any).createSession({
+        cwd: projectDir,
+        mcpServers: [],
+        _meta: { disableBuiltInTools: true },
+      });
+
+      expect(getCapturedOptions().permissionMode).toBe("auto");
+      expect(setPermissionModeSpy).not.toHaveBeenCalled();
+      expect(response.modes.currentModeId).toBe("auto");
+    });
  });

  it("resolves model aliases like opus[1m] to the correct model", async () => {
--- a/src/tests/acp-agent.test.ts
+++ b/src/tests/acp-agent.test.ts
--- a/src/tests/additional-roots.test.ts
+++ b/src/tests/additional-roots.test.ts
@@ -0,0 +1,79 @@
+import { mkdtemp, rm } from "node:fs/promises";
+import * as os from "node:os";
+import * as path from "node:path";
+import { AgentSideConnection, SessionNotification } from "@agentclientprotocol/sdk";
+import type { Options } from "@anthropic-ai/claude-agent-sdk";
+import type { ClaudeAcpAgent as ClaudeAcpAgentType } from "../acp-agent.js";
+import { afterEach, beforeEach, describe, expect, it, vi } from "vitest";
+
+let capturedOptions: Options | undefined;
+vi.mock("@anthropic-ai/claude-agent-sdk", async () => ({
+  ...(await vi.importActual<typeof import("@anthropic-ai/claude-agent-sdk")>(
+    "@anthropic-ai/claude-agent-sdk",
+  )),
+  query: ({ options }: { options: Options }) => {
+    capturedOptions = options;
+    return {
+      initializationResult: async () => ({
+        models: [
+          {
+            value: "claude-sonnet-4-6",
+            displayName: "Claude Sonnet",
+            description: "Fast",
+            supportsAutoMode: true,
+          },
+        ],
+      }),
+      setModel: async () => {},
+      setPermissionMode: async () => {},
+      supportedCommands: async () => [],
+      [Symbol.asyncIterator]: async function* () {},
+    };
+  },
+}));
+vi.mock("../tools.js", async () => ({
+  ...(await vi.importActual<typeof import("../tools.js")>("../tools.js")),
+  registerHookCallback: vi.fn(),
+}));
+
+describe("additionalRoots", () => {
+  let agent: ClaudeAcpAgentType;
+  const tempDirs: string[] = [];
+  const newSession = (meta: Record<string, unknown>, cwd = "/test") =>
+    agent.newSession({ cwd, mcpServers: [], _meta: meta });
+
+  beforeEach(async () => {
+    capturedOptions = undefined;
+    tempDirs.length = 0;
+    vi.resetModules();
+    const { ClaudeAcpAgent } = await import("../acp-agent.js");
+    agent = new ClaudeAcpAgent({
+      sessionUpdate: async (_notification: SessionNotification) => {},
+      requestPermission: async () => ({ outcome: { outcome: "cancelled" } }),
+      readTextFile: async () => ({ content: "" }),
+      writeTextFile: async () => ({}),
+    } as unknown as AgentSideConnection);
+  });
+
+  afterEach(
+    async () =>
+      void (await Promise.all(tempDirs.map((dir) => rm(dir, { recursive: true, force: true })))),
+  );
+
+  it("passes through relative roots as provided", async () => {
+    const projectRoot = await mkdtemp(path.join(os.tmpdir(), "claude-project-"));
+    tempDirs.push(projectRoot);
+    await newSession({ additionalRoots: ["."] }, projectRoot);
+    expect(capturedOptions!.additionalDirectories).toEqual(["."]);
+  });
+
+  it("merges additionalRoots with user additionalDirectories without normalization", async () => {
+    const root = await mkdtemp(path.join(os.tmpdir(), "claude-root-"));
+    tempDirs.push(root);
+    await newSession({
+      additionalRoots: ["", root],
+      claudeCode: { options: { additionalDirectories: ["/workspace/shared"] } },
+    });
+    expect(capturedOptions!.additionalDirectories).toEqual(["/workspace/shared", "", root]);
+  });
+});
--- a/src/tests/authorization.test.ts
+++ b/src/tests/authorization.test.ts
@@ -2,6 +2,23 @@ import { describe, expect, it, Mock, vi, afterEach, beforeEach } from "vitest";
 import { ClaudeAcpAgent } from "../acp-agent.js";
 import { AgentSideConnection } from "@agentclientprotocol/sdk";

+const mockQuery = vi.hoisted(() =>
+  vi.fn(() => ({
+    initializationResult: vi.fn().mockResolvedValue({
+      models: [
+        { value: "id", displayName: "name", description: "description", supportsAutoMode: true },
+      ],
+    }),
+    setModel: vi.fn(),
+    setPermissionMode: vi.fn(),
+    supportedCommands: vi.fn().mockResolvedValue([]),
+  })),
+);
+
+vi.mock("@anthropic-ai/claude-agent-sdk", () => ({
+  query: mockQuery,
+}));
+
 describe("authorization", () => {
  beforeEach(() => {
    vi.useFakeTimers();
@@ -17,20 +34,6 @@ describe("authorization", () => {
  });

  async function createAgentMock(): Promise<[ClaudeAcpAgent, Mock]> {
-    const mockQuery = vi.hoisted(() =>
-      vi.fn(() => ({
-        initializationResult: vi.fn().mockResolvedValue({
-          models: [{ value: "id", displayName: "name", description: "description" }],
-        }),
-        setModel: vi.fn(),
-        supportedCommands: vi.fn().mockResolvedValue([]),
-      })),
-    );
-
-    vi.mock("@anthropic-ai/claude-agent-sdk", () => ({
-      query: mockQuery,
-    }));
-
    const connectionMock = {
      sessionUpdate: async (_: any) => {},
    } as AgentSideConnection;
@@ -72,7 +75,7 @@ describe("authorization", () => {
    const initializeResponse = await agent.initialize({
      protocolVersion: 1,
      clientCapabilities: {
-        auth: { _meta: { gateway: true } },
+        auth: { terminal: true, _meta: { gateway: true } },
      } as any,
    });
    expect(initializeResponse.authMethods).toContainEqual(
@@ -123,14 +126,17 @@ describe("authorization", () => {
      } as any,
    });
    expect(initializeResponse.authMethods).not.toContainEqual(
-      expect.objectContaining({ id: "claude-login" }),
+      expect.objectContaining({ id: "claude-ai-login" }),
+    );
+    expect(initializeResponse.authMethods).not.toContainEqual(
+      expect.objectContaining({ id: "console-login" }),
    );
    expect(initializeResponse.authMethods).toContainEqual(
      expect.objectContaining({ id: "gateway" }),
    );
  });

-  it("terminal auth still offered when --hide-claude-auth is set", async () => {
+  it("hide claude auth but still show console login when terminal-auth is set", async () => {
    const [agent] = await createAgentMock();
    vi.stubGlobal("process", { ...process, argv: ["--hide-claude-auth"] });

@@ -140,7 +146,86 @@ describe("authorization", () => {
        _meta: { "terminal-auth": true },
      },
    });
+    expect(initializeResponse.authMethods).not.toContainEqual(
+      expect.objectContaining({ id: "claude-ai-login" }),
+    );
    expect(initializeResponse.authMethods).toContainEqual(
+      expect.objectContaining({ id: "console-login" }),
+    );
+  });
+
+  it("hide claude auth but still show console login with terminal capability", async () => {
+    const [agent] = await createAgentMock();
+    vi.stubGlobal("process", { ...process, argv: ["--hide-claude-auth"] });
+
+    const initializeResponse = await agent.initialize({
+      protocolVersion: 1,
+      clientCapabilities: {
+        auth: { terminal: true },
+      },
+    });
+    expect(initializeResponse.authMethods).not.toContainEqual(
+      expect.objectContaining({ id: "claude-ai-login" }),
+    );
+    expect(initializeResponse.authMethods).toContainEqual(
+      expect.objectContaining({ id: "console-login" }),
+    );
+  });
+
+  it("SSH session falls back to single legacy login method", async () => {
+    const [agent] = await createAgentMock();
+    vi.stubGlobal("process", { ...process, env: { ...process.env, SSH_TTY: "/dev/pts/0" } });
+
+    const initializeResponse = await agent.initialize({
+      protocolVersion: 1,
+      clientCapabilities: { auth: { terminal: true } },
+    });
+
+    expect(initializeResponse.authMethods).toContainEqual(
+      expect.objectContaining({ id: "claude-login" }),
+    );
+    expect(initializeResponse.authMethods).not.toContainEqual(
+      expect.objectContaining({ id: "claude-ai-login" }),
+    );
+    expect(initializeResponse.authMethods).not.toContainEqual(
+      expect.objectContaining({ id: "console-login" }),
+    );
+  });
+
+  it("CLAUDE_CODE_REMOTE falls back to single legacy login method", async () => {
+    const [agent] = await createAgentMock();
+    vi.stubGlobal("process", { ...process, env: { ...process.env, CLAUDE_CODE_REMOTE: "1" } });
+
+    const initializeResponse = await agent.initialize({
+      protocolVersion: 1,
+      clientCapabilities: { auth: { terminal: true } },
+    });
+
+    expect(initializeResponse.authMethods).toContainEqual(
+      expect.objectContaining({ id: "claude-login" }),
+    );
+    expect(initializeResponse.authMethods).not.toContainEqual(
+      expect.objectContaining({ id: "claude-ai-login" }),
+    );
+    expect(initializeResponse.authMethods).not.toContainEqual(
+      expect.objectContaining({ id: "console-login" }),
+    );
+  });
+
+  it("remote environment respects hide-claude-auth", async () => {
+    const [agent] = await createAgentMock();
+    vi.stubGlobal("process", {
+      ...process,
+      argv: ["--hide-claude-auth"],
+      env: { ...process.env, SSH_CONNECTION: "192.168.1.1 12345 192.168.1.2 22" },
+    });
+
+    const initializeResponse = await agent.initialize({
+      protocolVersion: 1,
+      clientCapabilities: { auth: { terminal: true } },
+    });
+
+    expect(initializeResponse.authMethods).not.toContainEqual(
      expect.objectContaining({ id: "claude-login" }),
    );
  });
@@ -150,11 +235,14 @@ describe("authorization", () => {

    const initializeResponse = await agent.initialize({
      protocolVersion: 1,
-      clientCapabilities: {},
+      clientCapabilities: { auth: { terminal: true } },
    });

    expect(initializeResponse.authMethods).toContainEqual(
-      expect.objectContaining({ id: "claude-login" }),
+      expect.objectContaining({ id: "claude-ai-login" }),
+    );
+    expect(initializeResponse.authMethods).toContainEqual(
+      expect.objectContaining({ id: "console-login" }),
    );
  });
 });
--- a/src/tests/create-session-options.test.ts
+++ b/src/tests/create-session-options.test.ts
@@ -1,4 +1,4 @@
-import { describe, it, expect, beforeEach, vi } from "vitest";
+import { describe, it, expect, beforeEach, afterEach, vi } from "vitest";
 import { AgentSideConnection, SessionNotification } from "@agentclientprotocol/sdk";
 import type { Options } from "@anthropic-ai/claude-agent-sdk";
 import type { ClaudeAcpAgent as ClaudeAcpAgentType } from "../acp-agent.js";
@@ -15,10 +15,16 @@ vi.mock("@anthropic-ai/claude-agent-sdk", async () => {
      return {
        initializationResult: async () => ({
          models: [
-            { value: "claude-sonnet-4-5", displayName: "Claude Sonnet", description: "Fast" },
+            {
+              value: "claude-sonnet-4-6",
+              displayName: "Claude Sonnet",
+              description: "Fast",
+              supportsAutoMode: true,
+            },
          ],
        }),
        setModel: async () => {},
+        setPermissionMode: async () => {},
        supportedCommands: async () => [],
        [Symbol.asyncIterator]: async function* () {},
      };
@@ -102,7 +108,7 @@ describe("createSession options merging", () => {
    expect(capturedOptions!.disallowedTools).toContain("AskUserQuestion");
  });

-  it("includes both user and built-in disallowed tools when disableBuiltInTools is true", async () => {
+  it("sets tools to empty array when disableBuiltInTools is true", async () => {
    await agent.newSession({
      cwd: "/test",
      mcpServers: [],
@@ -116,15 +122,11 @@ describe("createSession options merging", () => {
      },
    });

-    const disallowed = capturedOptions!.disallowedTools!;
-    // User-provided
-    expect(disallowed).toContain("CustomTool");
-    // ACP internal
-    expect(disallowed).toContain("AskUserQuestion");
-    // Built-in tools disabled by disableBuiltInTools
-    expect(disallowed).toContain("Read");
-    expect(disallowed).toContain("Write");
-    expect(disallowed).toContain("Bash");
+    // disableBuiltInTools removes all built-in tools from context
+    expect(capturedOptions!.tools).toEqual([]);
+    // User-provided and ACP disallowedTools still apply
+    expect(capturedOptions!.disallowedTools).toContain("CustomTool");
+    expect(capturedOptions!.disallowedTools).toContain("AskUserQuestion");
  });

  it("merges user-provided hooks with ACP hooks", async () => {
@@ -199,6 +201,247 @@ describe("createSession options merging", () => {
    expect(capturedOptions?.env?.HOME).toBe("/custom/home");
  });

+  it("defaults tools to claude_code preset when not provided", async () => {
+    await agent.newSession({
+      cwd: "/test",
+      mcpServers: [],
+    });
+
+    expect(capturedOptions!.tools).toEqual({ type: "preset", preset: "claude_code" });
+  });
+
+  it("passes through user-provided tools string array", async () => {
+    await agent.newSession({
+      cwd: "/test",
+      mcpServers: [],
+      _meta: {
+        claudeCode: {
+          options: {
+            tools: ["Read", "Glob"],
+          },
+        },
+      },
+    });
+
+    expect(capturedOptions!.tools).toEqual(["Read", "Glob"]);
+  });
+
+  it("explicit tools array takes precedence over disableBuiltInTools", async () => {
+    await agent.newSession({
+      cwd: "/test",
+      mcpServers: [],
+      _meta: {
+        disableBuiltInTools: true,
+        claudeCode: {
+          options: {
+            tools: ["Read", "Glob"],
+          },
+        },
+      },
+    });
+
+    expect(capturedOptions!.tools).toEqual(["Read", "Glob"]);
+  });
+
+  it("passes through empty tools array to disable all built-in tools", async () => {
+    await agent.newSession({
+      cwd: "/test",
+      mcpServers: [],
+      _meta: {
+        claudeCode: {
+          options: {
+            tools: [],
+          },
+        },
+      },
+    });
+
+    expect(capturedOptions!.tools).toEqual([]);
+  });
+
+  describe("systemPrompt via _meta", () => {
+    it("defaults to the claude_code preset when not provided", async () => {
+      await agent.newSession({ cwd: "/test", mcpServers: [] });
+
+      expect(capturedOptions!.systemPrompt).toEqual({
+        type: "preset",
+        preset: "claude_code",
+      });
+    });
+
+    it("replaces the preset when a string is provided", async () => {
+      await agent.newSession({
+        cwd: "/test",
+        mcpServers: [],
+        _meta: { systemPrompt: "custom prompt" },
+      });
+
+      expect(capturedOptions!.systemPrompt).toBe("custom prompt");
+    });
+
+    it("forwards append", async () => {
+      await agent.newSession({
+        cwd: "/test",
+        mcpServers: [],
+        _meta: { systemPrompt: { append: "extra instructions" } },
+      });
+
+      expect(capturedOptions!.systemPrompt).toEqual({
+        type: "preset",
+        preset: "claude_code",
+        append: "extra instructions",
+      });
+    });
+
+    it("forwards excludeDynamicSections", async () => {
+      await agent.newSession({
+        cwd: "/test",
+        mcpServers: [],
+        _meta: { systemPrompt: { excludeDynamicSections: true } },
+      });
+
+      expect(capturedOptions!.systemPrompt).toEqual({
+        type: "preset",
+        preset: "claude_code",
+        excludeDynamicSections: true,
+      });
+    });
+
+    it("forwards append and excludeDynamicSections together", async () => {
+      await agent.newSession({
+        cwd: "/test",
+        mcpServers: [],
+        _meta: {
+          systemPrompt: {
+            append: "extra instructions",
+            excludeDynamicSections: true,
+          },
+        },
+      });
+
+      expect(capturedOptions!.systemPrompt).toEqual({
+        type: "preset",
+        preset: "claude_code",
+        append: "extra instructions",
+        excludeDynamicSections: true,
+      });
+    });
+
+    it("ignores caller-provided type/preset overrides", async () => {
+      await agent.newSession({
+        cwd: "/test",
+        mcpServers: [],
+        _meta: {
+          systemPrompt: {
+            type: "something-else",
+            preset: "other-preset",
+            append: "extra",
+          },
+        },
+      });
+
+      expect(capturedOptions!.systemPrompt).toEqual({
+        type: "preset",
+        preset: "claude_code",
+        append: "extra",
+      });
+    });
+  });
+
+  describe("CLAUDE_MODEL_CONFIG", () => {
+    let originalModelConfig: string | undefined;
+
+    beforeEach(() => {
+      originalModelConfig = process.env.CLAUDE_MODEL_CONFIG;
+      delete process.env.CLAUDE_MODEL_CONFIG;
+    });
+
+    afterEach(() => {
+      if (originalModelConfig !== undefined) {
+        process.env.CLAUDE_MODEL_CONFIG = originalModelConfig;
+      } else {
+        delete process.env.CLAUDE_MODEL_CONFIG;
+      }
+    });
+
+    it("passes modelOverrides as settings", async () => {
+      process.env.CLAUDE_MODEL_CONFIG = JSON.stringify({
+        modelOverrides: { "claude-opus-4-6": "us.anthropic.claude-opus-4-6-v1" },
+      });
+
+      await agent.newSession({ cwd: "/test", mcpServers: [] });
+
+      expect(capturedOptions!.settings).toEqual({
+        modelOverrides: { "claude-opus-4-6": "us.anthropic.claude-opus-4-6-v1" },
+      });
+    });
+
+    it("passes availableModels as settings", async () => {
+      process.env.CLAUDE_MODEL_CONFIG = JSON.stringify({
+        availableModels: ["opus", "sonnet"],
+      });
+
+      await agent.newSession({ cwd: "/test", mcpServers: [] });
+
+      expect(capturedOptions!.settings).toEqual({
+        availableModels: ["opus", "sonnet"],
+      });
+    });
+
+    it("passes both modelOverrides and availableModels", async () => {
+      process.env.CLAUDE_MODEL_CONFIG = JSON.stringify({
+        modelOverrides: { "claude-opus-4-6": "us.anthropic.claude-opus-4-6-v1" },
+        availableModels: ["opus"],
+      });
+
+      await agent.newSession({ cwd: "/test", mcpServers: [] });
+
+      expect(capturedOptions!.settings).toEqual({
+        modelOverrides: { "claude-opus-4-6": "us.anthropic.claude-opus-4-6-v1" },
+        availableModels: ["opus"],
+      });
+    });
+
+    it("does not add settings when env var is not set", async () => {
+      await agent.newSession({ cwd: "/test", mcpServers: [] });
+
+      expect(capturedOptions!.settings).toBeUndefined();
+    });
+
+    it("ignores env var when _meta provides settings", async () => {
+      process.env.CLAUDE_MODEL_CONFIG = JSON.stringify({
+        modelOverrides: { "claude-opus-4-6": "us.anthropic.claude-opus-4-6-v1" },
+      });
+
+      await agent.newSession({
+        cwd: "/test",
+        mcpServers: [],
+        _meta: {
+          claudeCode: {
+            options: {
+              settings: {
+                model: "claude-sonnet-4-6",
+                modelOverrides: { "claude-opus-4-6": "meta-value" },
+              },
+            },
+          },
+        },
+      });
+
+      // _meta settings take precedence; env var is ignored entirely
+      expect(capturedOptions!.settings).toEqual({
+        model: "claude-sonnet-4-6",
+        modelOverrides: { "claude-opus-4-6": "meta-value" },
+      });
+    });
+
+    it("throws on invalid JSON", async () => {
+      process.env.CLAUDE_MODEL_CONFIG = "not-json";
+
+      await expect(agent.newSession({ cwd: "/test", mcpServers: [] })).rejects.toThrow();
+    });
+  });
+
  it("merges user-provided mcpServers with ACP mcpServers", async () => {
    await agent.newSession({
      cwd: "/test",
--- a/src/tests/resolve-permission-mode.test.ts
+++ b/src/tests/resolve-permission-mode.test.ts
@@ -1,5 +1,12 @@
-import { describe, it, expect } from "vitest";
-import { resolvePermissionMode } from "../acp-agent.js";
+import { describe, it, expect, vi } from "vitest";
+import { resolvePermissionMode, type Logger } from "../acp-agent.js";
+
+function mockLogger() {
+  const error = vi.fn<(...args: any[]) => void>();
+  const log = vi.fn<(...args: any[]) => void>();
+  const logger: Logger = { log, error };
+  return { logger, error, log };
+}

 describe("resolvePermissionMode", () => {
  it("returns 'default' when no mode is provided", () => {
@@ -26,18 +33,25 @@ describe("resolvePermissionMode", () => {
    expect(resolvePermissionMode("  dontAsk  ")).toBe("dontAsk");
  });

-  it("throws on non-string values", () => {
-    expect(() => resolvePermissionMode(123)).toThrow("expected a string");
-    expect(() => resolvePermissionMode(true)).toThrow("expected a string");
-    expect(() => resolvePermissionMode({})).toThrow("expected a string");
+  it("falls back to 'default' and logs on non-string values", () => {
+    for (const value of [123, true, {}]) {
+      const { logger, error } = mockLogger();
+      expect(resolvePermissionMode(value, logger)).toBe("default");
+      expect(error).toHaveBeenCalledWith(expect.stringContaining("expected a string"));
+    }
  });

-  it("throws on empty string", () => {
-    expect(() => resolvePermissionMode("")).toThrow("expected a non-empty string");
-    expect(() => resolvePermissionMode("  ")).toThrow("expected a non-empty string");
+  it("falls back to 'default' and logs on empty string", () => {
+    for (const value of ["", "  "]) {
+      const { logger, error } = mockLogger();
+      expect(resolvePermissionMode(value, logger)).toBe("default");
+      expect(error).toHaveBeenCalledWith(expect.stringContaining("expected a non-empty string"));
+    }
  });

-  it("throws on unknown mode", () => {
-    expect(() => resolvePermissionMode("yolo")).toThrow("Invalid permissions.defaultMode: yolo");
+  it("falls back to 'default' and logs on unknown mode", () => {
+    const { logger, error } = mockLogger();
+    expect(resolvePermissionMode("yolo", logger)).toBe("default");
+    expect(error).toHaveBeenCalledWith(expect.stringContaining("yolo"));
  });
 });
--- a/src/tests/session-config-options.test.ts
+++ b/src/tests/session-config-options.test.ts
@@ -1,5 +1,6 @@
 import { describe, it, expect, beforeEach, vi } from "vitest";
 import { AgentSideConnection, SessionNotification } from "@agentclientprotocol/sdk";
+import type { ModelInfo } from "@anthropic-ai/claude-agent-sdk";
 import type { ClaudeAcpAgent as ClaudeAcpAgentType } from "../acp-agent.js";

 const { registerHookCallbackSpy } = vi.hoisted(() => ({
@@ -29,7 +30,7 @@ const MOCK_MODELS = {
  currentModelId: "claude-opus-4-5",
  availableModels: [
    { modelId: "claude-opus-4-5", name: "Claude Opus", description: "Most capable" },
-    { modelId: "claude-sonnet-4-5", name: "Claude Sonnet", description: "Balanced" },
+    { modelId: "claude-sonnet-4-6", name: "Claude Sonnet", description: "Balanced" },
  ],
 };

@@ -58,6 +59,19 @@ const MOCK_CONFIG_OPTIONS = [
      description: m.description,
    })),
  },
+  {
+    id: "effort",
+    name: "Effort",
+    description: "Available effort levels for this model",
+    type: "select",
+    category: "effort",
+    currentValue: "high",
+    options: [
+      { value: "low", name: "Low" },
+      { value: "medium", name: "Medium" },
+      { value: "high", name: "High" },
+    ],
+  },
 ];

 describe("session config options", () => {
@@ -67,6 +81,7 @@ describe("session config options", () => {
  let createSessionSpy: ReturnType<typeof vi.fn>;
  let setPermissionModeSpy: ReturnType<typeof vi.fn>;
  let setModelSpy: ReturnType<typeof vi.fn>;
+  let applyFlagSettingsSpy: ReturnType<typeof vi.fn>;

  function createMockClient(): AgentSideConnection {
    return {
@@ -82,18 +97,32 @@ describe("session config options", () => {
  function populateSession() {
    setPermissionModeSpy = vi.fn();
    setModelSpy = vi.fn();
+    applyFlagSettingsSpy = vi.fn();

    (agent as unknown as { sessions: Record<string, unknown> }).sessions[SESSION_ID] = {
      query: {
        setPermissionMode: setPermissionModeSpy,
        setModel: setModelSpy,
+        applyFlagSettings: applyFlagSettingsSpy,
        supportedCommands: async () => [],
      },
      input: null,
      cancelled: false,
      permissionMode: "default",
      settingsManager: {},
+      modes: structuredClone(MOCK_MODES),
+      models: structuredClone(MOCK_MODELS),
+      modelInfos: MOCK_MODELS.availableModels.map(
+        (m): ModelInfo => ({
+          value: m.modelId,
+          displayName: m.name,
+          description: m.description,
+          supportsEffort: true,
+          supportedEffortLevels: ["low", "medium", "high"],
+        }),
+      ),
      configOptions: structuredClone(MOCK_CONFIG_OPTIONS),
+      contextWindowSize: 200000,
    };
  }

@@ -213,10 +242,10 @@ describe("session config options", () => {
      await agent.setSessionConfigOption({
        sessionId: SESSION_ID,
        configId: "model",
-        value: "claude-sonnet-4-5",
+        value: "claude-sonnet-4-6",
      });

-      expect(setModelSpy).toHaveBeenCalledWith("claude-sonnet-4-5");
+      expect(setModelSpy).toHaveBeenCalledWith("claude-sonnet-4-6");

      const configUpdate = sessionUpdates.find(
        (n) => n.update.sessionUpdate === "config_option_update",
@@ -224,6 +253,61 @@ describe("session config options", () => {
      expect(configUpdate).toBeUndefined();
    });

+    it("resolves model alias 'opus' to full model ID", async () => {
+      const response = await agent.setSessionConfigOption({
+        sessionId: SESSION_ID,
+        configId: "model",
+        value: "opus",
+      });
+
+      expect(setModelSpy).toHaveBeenCalledWith("claude-opus-4-5");
+
+      const modelOption = response.configOptions.find((o) => o.id === "model");
+      expect(modelOption?.currentValue).toBe("claude-opus-4-5");
+    });
+
+    it("resolves model alias 'sonnet' to full model ID", async () => {
+      await agent.setSessionConfigOption({
+        sessionId: SESSION_ID,
+        configId: "model",
+        value: "sonnet",
+      });
+
+      expect(setModelSpy).toHaveBeenCalledWith("claude-sonnet-4-6");
+    });
+
+    it("resolves display name to model ID", async () => {
+      await agent.setSessionConfigOption({
+        sessionId: SESSION_ID,
+        configId: "model",
+        value: "Claude Sonnet",
+      });
+
+      expect(setModelSpy).toHaveBeenCalledWith("claude-sonnet-4-6");
+    });
+
+    it("still works with exact model ID", async () => {
+      const response = await agent.setSessionConfigOption({
+        sessionId: SESSION_ID,
+        configId: "model",
+        value: "claude-sonnet-4-6",
+      });
+
+      expect(setModelSpy).toHaveBeenCalledWith("claude-sonnet-4-6");
+      const modelOption = response.configOptions.find((o) => o.id === "model");
+      expect(modelOption?.currentValue).toBe("claude-sonnet-4-6");
+    });
+
+    it("throws for completely invalid model value", async () => {
+      await expect(
+        agent.setSessionConfigOption({
+          sessionId: SESSION_ID,
+          configId: "model",
+          value: "gpt-4",
+        }),
+      ).rejects.toThrow("Invalid value for config option model: gpt-4");
+    });
+
    it("returns full configOptions in the response", async () => {
      const response = await agent.setSessionConfigOption({
        sessionId: SESSION_ID,
@@ -300,7 +384,7 @@ describe("session config options", () => {
    it("sends config_option_update when model is changed via setSessionModel", async () => {
      await agent.unstable_setSessionModel({
        sessionId: SESSION_ID,
-        modelId: "claude-sonnet-4-5",
+        modelId: "claude-sonnet-4-6",
      });

      const configUpdate = sessionUpdates.find(
@@ -310,7 +394,7 @@ describe("session config options", () => {
      expect(configUpdate?.update).toMatchObject({
        sessionUpdate: "config_option_update",
        configOptions: expect.arrayContaining([
-          expect.objectContaining({ id: "model", currentValue: "claude-sonnet-4-5" }),
+          expect.objectContaining({ id: "model", currentValue: "claude-sonnet-4-6" }),
        ]),
      });
    });
@@ -318,7 +402,7 @@ describe("session config options", () => {
    it("updates stored configOptions currentValue when model changes", async () => {
      await agent.unstable_setSessionModel({
        sessionId: SESSION_ID,
-        modelId: "claude-sonnet-4-5",
+        modelId: "claude-sonnet-4-6",
      });

      const session = (
@@ -327,7 +411,102 @@ describe("session config options", () => {
        }
      ).sessions[SESSION_ID];
      const modelOption = session.configOptions.find((o) => o.id === "model");
-      expect(modelOption?.currentValue).toBe("claude-sonnet-4-5");
+      expect(modelOption?.currentValue).toBe("claude-sonnet-4-6");
+    });
+
+    it("includes updated effort in config_option_update when model drops effort support", async () => {
+      const session = (agent as unknown as { sessions: Record<string, any> }).sessions[SESSION_ID];
+      session.modelInfos = [
+        {
+          value: "claude-opus-4-5",
+          displayName: "Claude Opus",
+          description: "Most capable",
+          supportsEffort: true,
+          supportedEffortLevels: ["low", "medium", "high"],
+        },
+        {
+          value: "claude-sonnet-4-6",
+          displayName: "Claude Sonnet",
+          description: "Balanced",
+          supportsEffort: false,
+        },
+      ];
+
+      await agent.unstable_setSessionModel({
+        sessionId: SESSION_ID,
+        modelId: "claude-sonnet-4-6",
+      });
+
+      const configUpdate = sessionUpdates.find(
+        (n) => n.update.sessionUpdate === "config_option_update",
+      );
+      expect(configUpdate).toBeDefined();
+      const effortOption = (configUpdate?.update as any).configOptions.find(
+        (o: any) => o.id === "effort",
+      );
+      expect(effortOption).toBeUndefined();
+      expect(applyFlagSettingsSpy).toHaveBeenCalledWith({ effortLevel: undefined });
+    });
+
+    it("clamps effort in config_option_update when new model has different supported levels", async () => {
+      // Set current effort to "max" which the new model won't support
+      const session = (agent as unknown as { sessions: Record<string, any> }).sessions[SESSION_ID];
+      const effortOpt = session.configOptions.find((o: any) => o.id === "effort");
+      if (effortOpt) effortOpt.currentValue = "max";
+
+      session.modelInfos = [
+        {
+          value: "claude-opus-4-5",
+          displayName: "Claude Opus",
+          description: "Most capable",
+          supportsEffort: true,
+          supportedEffortLevels: ["low", "medium", "high", "max"],
+        },
+        {
+          value: "claude-sonnet-4-6",
+          displayName: "Claude Sonnet",
+          description: "Balanced",
+          supportsEffort: true,
+          supportedEffortLevels: ["low", "medium", "high"],
+        },
+      ];
+
+      await agent.unstable_setSessionModel({
+        sessionId: SESSION_ID,
+        modelId: "claude-sonnet-4-6",
+      });
+
+      const configUpdate = sessionUpdates.find(
+        (n) => n.update.sessionUpdate === "config_option_update",
+      );
+      const effortOption = (configUpdate?.update as any).configOptions.find(
+        (o: any) => o.id === "effort",
+      );
+      expect(effortOption).toBeDefined();
+      expect(effortOption.currentValue).toBe("high");
+      expect(applyFlagSettingsSpy).toHaveBeenCalledWith({ effortLevel: "high" });
+    });
+
+    it("preserves effort in config_option_update when new model supports same level", async () => {
+      // Set effort to "low" first
+      const session = (agent as unknown as { sessions: Record<string, any> }).sessions[SESSION_ID];
+      const effortOpt = session.configOptions.find((o: any) => o.id === "effort");
+      if (effortOpt) effortOpt.currentValue = "low";
+
+      await agent.unstable_setSessionModel({
+        sessionId: SESSION_ID,
+        modelId: "claude-sonnet-4-6",
+      });
+
+      const configUpdate = sessionUpdates.find(
+        (n) => n.update.sessionUpdate === "config_option_update",
+      );
+      const effortOption = (configUpdate?.update as any).configOptions.find(
+        (o: any) => o.id === "effort",
+      );
+      expect(effortOption?.currentValue).toBe("low");
+      // Effort didn't change, so applyFlagSettings should NOT be called
+      expect(applyFlagSettingsSpy).not.toHaveBeenCalled();
    });
  });

@@ -353,7 +532,7 @@ describe("session config options", () => {
      await agent.setSessionConfigOption({
        sessionId: SESSION_ID,
        configId: "model",
-        value: "claude-sonnet-4-5",
+        value: "claude-sonnet-4-6",
      });

      const configUpdates = sessionUpdates.filter(
@@ -363,6 +542,223 @@ describe("session config options", () => {
    });
  });

+  describe("setSessionConfigOption for effort", () => {
+    beforeEach(() => {
+      populateSession();
+    });
+
+    it("calls applyFlagSettings with effortLevel", async () => {
+      await agent.setSessionConfigOption({
+        sessionId: SESSION_ID,
+        configId: "effort",
+        value: "low",
+      });
+
+      expect(applyFlagSettingsSpy).toHaveBeenCalledWith({ effortLevel: "low" });
+    });
+
+    it("updates effort currentValue in returned configOptions", async () => {
+      const response = await agent.setSessionConfigOption({
+        sessionId: SESSION_ID,
+        configId: "effort",
+        value: "medium",
+      });
+
+      const effortOption = response.configOptions.find((o) => o.id === "effort");
+      expect(effortOption?.currentValue).toBe("medium");
+    });
+
+    it("throws for invalid effort value", async () => {
+      await expect(
+        agent.setSessionConfigOption({
+          sessionId: SESSION_ID,
+          configId: "effort",
+          value: "turbo",
+        }),
+      ).rejects.toThrow("Invalid value for config option effort: turbo");
+    });
+
+    it("does not send config_option_update notification", async () => {
+      await agent.setSessionConfigOption({
+        sessionId: SESSION_ID,
+        configId: "effort",
+        value: "low",
+      });
+
+      const configUpdates = sessionUpdates.filter(
+        (n) => n.update.sessionUpdate === "config_option_update",
+      );
+      expect(configUpdates).toHaveLength(0);
+    });
+
+    it("other options are unchanged when effort is updated", async () => {
+      const response = await agent.setSessionConfigOption({
+        sessionId: SESSION_ID,
+        configId: "effort",
+        value: "low",
+      });
+
+      const modeOption = response.configOptions.find((o) => o.id === "mode");
+      expect(modeOption?.currentValue).toBe("default");
+      const modelOption = response.configOptions.find((o) => o.id === "model");
+      expect(modelOption?.currentValue).toBe("claude-opus-4-5");
+    });
+  });
+
+  describe("effort level and model switch interactions", () => {
+    beforeEach(() => {
+      populateSession();
+    });
+
+    it("drops effort option when switching to a model without effort support", async () => {
+      // Make sonnet not support effort
+      const session = (agent as unknown as { sessions: Record<string, any> }).sessions[SESSION_ID];
+      session.modelInfos = [
+        {
+          value: "claude-opus-4-5",
+          displayName: "Claude Opus",
+          description: "Most capable",
+          supportsEffort: true,
+          supportedEffortLevels: ["low", "medium", "high"],
+        },
+        {
+          value: "claude-sonnet-4-6",
+          displayName: "Claude Sonnet",
+          description: "Balanced",
+          supportsEffort: false,
+        },
+      ];
+
+      const response = await agent.setSessionConfigOption({
+        sessionId: SESSION_ID,
+        configId: "model",
+        value: "claude-sonnet-4-6",
+      });
+
+      const effortOption = response.configOptions.find((o) => o.id === "effort");
+      expect(effortOption).toBeUndefined();
+    });
+
+    it("clears effort via applyFlagSettings when switching to a model without effort", async () => {
+      const session = (agent as unknown as { sessions: Record<string, any> }).sessions[SESSION_ID];
+      session.modelInfos = [
+        {
+          value: "claude-opus-4-5",
+          displayName: "Claude Opus",
+          description: "Most capable",
+          supportsEffort: true,
+          supportedEffortLevels: ["low", "medium", "high"],
+        },
+        {
+          value: "claude-sonnet-4-6",
+          displayName: "Claude Sonnet",
+          description: "Balanced",
+          supportsEffort: false,
+        },
+      ];
+
+      await agent.setSessionConfigOption({
+        sessionId: SESSION_ID,
+        configId: "model",
+        value: "claude-sonnet-4-6",
+      });
+
+      expect(applyFlagSettingsSpy).toHaveBeenCalledWith({ effortLevel: undefined });
+    });
+
+    it("adds effort option when switching to a model that supports effort", async () => {
+      const session = (agent as unknown as { sessions: Record<string, any> }).sessions[SESSION_ID];
+      // Start with sonnet (no effort) as current
+      session.models = { ...session.models, currentModelId: "claude-sonnet-4-6" };
+      session.modelInfos = [
+        {
+          value: "claude-opus-4-5",
+          displayName: "Claude Opus",
+          description: "Most capable",
+          supportsEffort: true,
+          supportedEffortLevels: ["low", "medium", "high"],
+        },
+        {
+          value: "claude-sonnet-4-6",
+          displayName: "Claude Sonnet",
+          description: "Balanced",
+          supportsEffort: false,
+        },
+      ];
+      // Remove effort from current config options
+      session.configOptions = session.configOptions.filter((o: any) => o.id !== "effort");
+
+      const response = await agent.setSessionConfigOption({
+        sessionId: SESSION_ID,
+        configId: "model",
+        value: "claude-opus-4-5",
+      });
+
+      const effortOption = response.configOptions.find((o) => o.id === "effort");
+      expect(effortOption).toBeDefined();
+      // No previous effort, so defaults to "high" (the API default)
+      expect(effortOption?.currentValue).toBe("high");
+    });
+
+    it("clamps effort to valid value when new model has different supported levels", async () => {
+      const session = (agent as unknown as { sessions: Record<string, any> }).sessions[SESSION_ID];
+      // Set current effort to "max" (not supported by sonnet in our mock)
+      const effortOpt = session.configOptions.find((o: any) => o.id === "effort");
+      if (effortOpt) effortOpt.currentValue = "max";
+
+      session.modelInfos = [
+        {
+          value: "claude-opus-4-5",
+          displayName: "Claude Opus",
+          description: "Most capable",
+          supportsEffort: true,
+          supportedEffortLevels: ["low", "medium", "high", "max"],
+        },
+        {
+          value: "claude-sonnet-4-6",
+          displayName: "Claude Sonnet",
+          description: "Balanced",
+          supportsEffort: true,
+          supportedEffortLevels: ["low", "medium", "high"],
+        },
+      ];
+
+      const response = await agent.setSessionConfigOption({
+        sessionId: SESSION_ID,
+        configId: "model",
+        value: "claude-sonnet-4-6",
+      });
+
+      const effortOption = response.configOptions.find((o) => o.id === "effort");
+      expect(effortOption).toBeDefined();
+      // "max" is not in sonnet's levels, so should fall back to "high" (the API default)
+      expect(effortOption?.currentValue).toBe("high");
+      // SDK should be told about the clamped value
+      expect(applyFlagSettingsSpy).toHaveBeenCalledWith({ effortLevel: "high" });
+    });
+
+    it("preserves effort value when new model supports the same level", async () => {
+      // Set effort to "low"
+      await agent.setSessionConfigOption({
+        sessionId: SESSION_ID,
+        configId: "effort",
+        value: "low",
+      });
+
+      // Switch model — both support "low"
+      const response = await agent.setSessionConfigOption({
+        sessionId: SESSION_ID,
+        configId: "model",
+        value: "claude-sonnet-4-6",
+      });
+
+      const effortOption = response.configOptions.find((o) => o.id === "effort");
+      expect(effortOption?.currentValue).toBe("low");
+      // applyFlagSettings was called once for the effort change, but not again for the model switch
+      expect(applyFlagSettingsSpy).toHaveBeenCalledTimes(1);
+    });
+  });
+
  describe("bidirectional consistency", () => {
    beforeEach(() => {
      populateSession();
@@ -382,10 +778,10 @@ describe("session config options", () => {
      await agent.setSessionConfigOption({
        sessionId: SESSION_ID,
        configId: "model",
-        value: "claude-sonnet-4-5",
+        value: "claude-sonnet-4-6",
      });

-      expect(setModelSpy).toHaveBeenCalledWith("claude-sonnet-4-5");
+      expect(setModelSpy).toHaveBeenCalledWith("claude-sonnet-4-6");
    });

    it("setSessionMode also syncs configOptions", async () => {
@@ -402,7 +798,7 @@ describe("session config options", () => {
    it("setSessionModel also syncs configOptions", async () => {
      await agent.unstable_setSessionModel({
        sessionId: SESSION_ID,
-        modelId: "claude-sonnet-4-5",
+        modelId: "claude-sonnet-4-6",
      });

      const session = (
@@ -411,8 +807,395 @@ describe("session config options", () => {
        }
      ).sessions[SESSION_ID];
      expect(session.configOptions.find((o) => o.id === "model")?.currentValue).toBe(
-        "claude-sonnet-4-5",
+        "claude-sonnet-4-6",
      );
    });
  });
+
+  describe("auto mode availability per model", () => {
+    /**
+     * Augment the session populated by `populateSession()` with a Haiku entry
+     * (no `supportsAutoMode`), Opus + Sonnet entries with `supportsAutoMode:
+     * true`, and seed `availableModes` so it currently includes `auto`. This
+     * exercises the per-model recomputation done by `applyConfigOptionValue`
+     * on a model switch.
+     */
+    function setupHaikuOpusSession(currentModeId: string = "default") {
+      const session = (agent as unknown as { sessions: Record<string, any> }).sessions[SESSION_ID];
+      session.modelInfos = [
+        {
+          value: "claude-opus-4-5",
+          displayName: "Claude Opus",
+          description: "Most capable",
+          supportsEffort: true,
+          supportedEffortLevels: ["low", "medium", "high"],
+          supportsAutoMode: true,
+        },
+        {
+          value: "claude-sonnet-4-6",
+          displayName: "Claude Sonnet",
+          description: "Balanced",
+          supportsEffort: true,
+          supportedEffortLevels: ["low", "medium", "high"],
+          supportsAutoMode: true,
+        },
+        {
+          value: "claude-haiku-4-5",
+          displayName: "Claude Haiku",
+          description: "Fast",
+          supportsEffort: true,
+          supportedEffortLevels: ["low", "medium", "high"],
+          // supportsAutoMode intentionally omitted
+        },
+      ];
+      session.models = {
+        currentModelId: "claude-opus-4-5",
+        availableModels: [
+          { modelId: "claude-opus-4-5", name: "Claude Opus", description: "Most capable" },
+          { modelId: "claude-sonnet-4-6", name: "Claude Sonnet", description: "Balanced" },
+          { modelId: "claude-haiku-4-5", name: "Claude Haiku", description: "Fast" },
+        ],
+      };
+      session.modes = {
+        currentModeId,
+        availableModes: [
+          {
+            id: "auto",
+            name: "Auto",
+            description: "Use a model classifier to approve/deny permission prompts",
+          },
+          {
+            id: "default",
+            name: "Default",
+            description: "Standard behavior, prompts for dangerous operations",
+          },
+          {
+            id: "acceptEdits",
+            name: "Accept Edits",
+            description: "Auto-accept file edit operations",
+          },
+          { id: "plan", name: "Plan Mode", description: "Planning mode" },
+          {
+            id: "dontAsk",
+            name: "Don't Ask",
+            description: "Don't prompt for permissions, deny if not pre-approved",
+          },
+        ],
+      };
+      // Reflect the seeded availableModes/availableModels in configOptions so
+      // the pre-state matches what `createSession` would have produced for
+      // Opus, and `setSessionConfigOption` validation can accept the seeded
+      // model ids (notably the new Haiku entry).
+      session.configOptions = session.configOptions.map((o: any) => {
+        if (o.id === "mode") {
+          return {
+            ...o,
+            currentValue: currentModeId,
+            options: session.modes.availableModes.map((m: any) => ({
+              value: m.id,
+              name: m.name,
+              description: m.description,
+            })),
+          };
+        }
+        if (o.id === "model") {
+          return {
+            ...o,
+            currentValue: session.models.currentModelId,
+            options: session.models.availableModels.map((m: any) => ({
+              value: m.modelId,
+              name: m.name,
+              description: m.description,
+            })),
+          };
+        }
+        return o;
+      });
+      return session;
+    }
+
+    beforeEach(() => {
+      populateSession();
+    });
+
+    it("drops `auto` from available modes when switching to Haiku", async () => {
+      setupHaikuOpusSession("default");
+
+      await agent.unstable_setSessionModel({
+        sessionId: SESSION_ID,
+        modelId: "claude-haiku-4-5",
+      });
+
+      const configUpdate = sessionUpdates.find(
+        (n) => n.update.sessionUpdate === "config_option_update",
+      );
+      expect(configUpdate).toBeDefined();
+      const modeOption = (configUpdate?.update as any).configOptions.find(
+        (o: any) => o.id === "mode",
+      );
+      expect(modeOption).toBeDefined();
+      const modeValues = modeOption.options.map((o: any) => o.value);
+      expect(modeValues).not.toContain("auto");
+      expect(modeValues).toEqual(
+        expect.arrayContaining(["default", "acceptEdits", "plan", "dontAsk"]),
+      );
+    });
+
+    it("clamps to `default` and emits current_mode_update when Opus(auto) → Haiku", async () => {
+      setupHaikuOpusSession("auto");
+
+      await agent.unstable_setSessionModel({
+        sessionId: SESSION_ID,
+        modelId: "claude-haiku-4-5",
+      });
+
+      // SDK was synced to "default".
+      expect(setPermissionModeSpy).toHaveBeenCalledWith("default");
+
+      // current_mode_update was emitted before config_option_update so a
+      // client applying notifications in order observes the mode change
+      // before re-rendering the config-option list.
+      const modeUpdateIdx = sessionUpdates.findIndex(
+        (n) => n.update.sessionUpdate === "current_mode_update",
+      );
+      const configUpdateIdx = sessionUpdates.findIndex(
+        (n) => n.update.sessionUpdate === "config_option_update",
+      );
+      expect(modeUpdateIdx).toBeGreaterThanOrEqual(0);
+      expect(configUpdateIdx).toBeGreaterThanOrEqual(0);
+      expect(modeUpdateIdx).toBeLessThan(configUpdateIdx);
+      expect((sessionUpdates[modeUpdateIdx].update as any).currentModeId).toBe("default");
+
+      // configOptions reflect the clamped mode.
+      const modeOption = (sessionUpdates[configUpdateIdx].update as any).configOptions.find(
+        (o: any) => o.id === "mode",
+      );
+      expect(modeOption.currentValue).toBe("default");
+    });
+
+    it("re-adds `auto` when switching from Haiku back to Opus", async () => {
+      const session = setupHaikuOpusSession("default");
+      // Pretend Haiku is the current model with no `auto`.
+      session.models.currentModelId = "claude-haiku-4-5";
+      session.modes.availableModes = session.modes.availableModes.filter(
+        (m: any) => m.id !== "auto",
+      );
+      const modeOpt = session.configOptions.find((o: any) => o.id === "mode");
+      modeOpt.options = session.modes.availableModes.map((m: any) => ({
+        value: m.id,
+        name: m.name,
+        description: m.description,
+      }));
+
+      await agent.unstable_setSessionModel({
+        sessionId: SESSION_ID,
+        modelId: "claude-opus-4-5",
+      });
+
+      const configUpdate = sessionUpdates.find(
+        (n) => n.update.sessionUpdate === "config_option_update",
+      );
+      expect(configUpdate).toBeDefined();
+      const modeOption = (configUpdate?.update as any).configOptions.find(
+        (o: any) => o.id === "mode",
+      );
+      const modeValues = modeOption.options.map((o: any) => o.value);
+      expect(modeValues).toContain("auto");
+
+      // The current mode ("default") is still valid on Opus, so no
+      // current_mode_update should have been emitted by the model switch.
+      const modeUpdates = sessionUpdates.filter(
+        (n) => n.update.sessionUpdate === "current_mode_update",
+      );
+      expect(modeUpdates).toHaveLength(0);
+    });
+
+    it("preserves the current mode when it remains valid after a model switch", async () => {
+      setupHaikuOpusSession("plan");
+
+      await agent.unstable_setSessionModel({
+        sessionId: SESSION_ID,
+        modelId: "claude-haiku-4-5",
+      });
+
+      // `plan` is in availableModes for both Opus and Haiku, so no clamp.
+      expect(setPermissionModeSpy).not.toHaveBeenCalledWith("default");
+
+      const modeUpdates = sessionUpdates.filter(
+        (n) => n.update.sessionUpdate === "current_mode_update",
+      );
+      expect(modeUpdates).toHaveLength(0);
+
+      const configUpdate = sessionUpdates.find(
+        (n) => n.update.sessionUpdate === "config_option_update",
+      );
+      const modeOption = (configUpdate?.update as any).configOptions.find(
+        (o: any) => o.id === "mode",
+      );
+      expect(modeOption.currentValue).toBe("plan");
+    });
+
+    it("clamps mode and emits current_mode_update via setSessionConfigOption(model)", async () => {
+      // Mirrors the unstable_setSessionModel(auto → Haiku) test, but goes
+      // through the request/response API. The `current_mode_update` side
+      // effect must still fire so clients learn about the clamp regardless of
+      // which entry point triggered the model switch.
+      setupHaikuOpusSession("auto");
+
+      const response = await agent.setSessionConfigOption({
+        sessionId: SESSION_ID,
+        configId: "model",
+        value: "claude-haiku-4-5",
+      });
+
+      expect(setPermissionModeSpy).toHaveBeenCalledWith("default");
+
+      const modeUpdates = sessionUpdates.filter(
+        (n) => n.update.sessionUpdate === "current_mode_update",
+      );
+      expect(modeUpdates).toHaveLength(1);
+      expect((modeUpdates[0].update as any).currentModeId).toBe("default");
+
+      // setSessionConfigOption is a request/response API: it returns the new
+      // configOptions in the response rather than emitting a
+      // config_option_update notification.
+      const configUpdates = sessionUpdates.filter(
+        (n) => n.update.sessionUpdate === "config_option_update",
+      );
+      expect(configUpdates).toHaveLength(0);
+
+      const modeOption = response.configOptions.find((o: any) => o.id === "mode");
+      expect(modeOption).toBeDefined();
+      expect((modeOption as any).currentValue).toBe("default");
+      expect((modeOption as any).options.map((o: any) => o.value)).not.toContain("auto");
+    });
+
+    it("rejects direct setSessionMode to `auto` when the active model does not offer it", async () => {
+      const session = setupHaikuOpusSession("default");
+      session.models.currentModelId = "claude-haiku-4-5";
+      session.modes.availableModes = session.modes.availableModes.filter(
+        (mode: any) => mode.id !== "auto",
+      );
+
+      await expect(agent.setSessionMode({ sessionId: SESSION_ID, modeId: "auto" })).rejects.toThrow(
+        "Mode auto is not available in this session",
+      );
+
+      expect(setPermissionModeSpy).not.toHaveBeenCalledWith("auto");
+      expect(sessionUpdates).toHaveLength(0);
+    });
+  });
+
+  describe("ExitPlanMode permission options filtered by availableModes", () => {
+    let capturedPermissionRequest: any;
+    let permissionResponse: any;
+
+    beforeEach(() => {
+      capturedPermissionRequest = null;
+      permissionResponse = { outcome: { outcome: "cancelled" } };
+      // Replace the default mock client with one that captures the
+      // requestPermission call so we can assert on the offered options.
+      (agent as any).client = {
+        sessionUpdate: async (notification: SessionNotification) => {
+          sessionUpdates.push(notification);
+        },
+        requestPermission: async (params: any) => {
+          capturedPermissionRequest = params;
+          return permissionResponse;
+        },
+        readTextFile: async () => ({ content: "" }),
+        writeTextFile: async () => ({}),
+      };
+      populateSession();
+    });
+
+    it("omits the `auto` option on a model without supportsAutoMode", async () => {
+      const session = (agent as unknown as { sessions: Record<string, any> }).sessions[SESSION_ID];
+      // Haiku-shaped session: availableModes does NOT include `auto`.
+      session.modes = {
+        currentModeId: "plan",
+        availableModes: [
+          { id: "default", name: "Default", description: "Standard" },
+          { id: "acceptEdits", name: "Accept Edits", description: "Auto-accept edits" },
+          { id: "plan", name: "Plan Mode", description: "Planning mode" },
+          { id: "dontAsk", name: "Don't Ask", description: "Deny if not pre-approved" },
+        ],
+      };
+
+      const canUseTool = (agent as any).canUseTool(SESSION_ID);
+      const signal = new AbortController().signal;
+      try {
+        await canUseTool(
+          "ExitPlanMode",
+          { plan: "do stuff" },
+          { signal, suggestions: undefined, toolUseID: "toolu_1" },
+        );
+      } catch {
+        // The mock client returns `cancelled`, which makes canUseTool throw.
+        // We only care about the captured requestPermission options.
+      }
+
+      expect(capturedPermissionRequest).not.toBeNull();
+      const optionIds = capturedPermissionRequest.options.map((o: any) => o.optionId);
+      expect(optionIds).not.toContain("auto");
+      expect(optionIds).toEqual(expect.arrayContaining(["default", "acceptEdits", "plan"]));
+    });
+
+    it("denies a selected `auto` option if the client did not receive that option", async () => {
+      const session = (agent as unknown as { sessions: Record<string, any> }).sessions[SESSION_ID];
+      session.modes = {
+        currentModeId: "plan",
+        availableModes: [
+          { id: "default", name: "Default", description: "Standard" },
+          { id: "acceptEdits", name: "Accept Edits", description: "Auto-accept edits" },
+          { id: "plan", name: "Plan Mode", description: "Planning mode" },
+          { id: "dontAsk", name: "Don't Ask", description: "Deny if not pre-approved" },
+        ],
+      };
+      permissionResponse = { outcome: { outcome: "selected", optionId: "auto" } };
+
+      const canUseTool = (agent as any).canUseTool(SESSION_ID);
+      const result = await canUseTool(
+        "ExitPlanMode",
+        { plan: "do stuff" },
+        { signal: new AbortController().signal, suggestions: undefined, toolUseID: "toolu_2" },
+      );
+
+      expect(capturedPermissionRequest).not.toBeNull();
+      const optionIds = capturedPermissionRequest.options.map((o: any) => o.optionId);
+      expect(optionIds).not.toContain("auto");
+      expect(result.behavior).toBe("deny");
+      expect(sessionUpdates).toHaveLength(0);
+    });
+
+    it("includes the `auto` option on a model with supportsAutoMode", async () => {
+      const session = (agent as unknown as { sessions: Record<string, any> }).sessions[SESSION_ID];
+      session.modes = {
+        currentModeId: "plan",
+        availableModes: [
+          { id: "auto", name: "Auto", description: "Use a model classifier" },
+          { id: "default", name: "Default", description: "Standard" },
+          { id: "acceptEdits", name: "Accept Edits", description: "Auto-accept edits" },
+          { id: "plan", name: "Plan Mode", description: "Planning mode" },
+          { id: "dontAsk", name: "Don't Ask", description: "Deny if not pre-approved" },
+        ],
+      };
+
+      const canUseTool = (agent as any).canUseTool(SESSION_ID);
+      const signal = new AbortController().signal;
+      try {
+        await canUseTool(
+          "ExitPlanMode",
+          { plan: "do stuff" },
+          { signal, suggestions: undefined, toolUseID: "toolu_3" },
+        );
+      } catch {
+        // mock returns cancelled
+      }
+
+      expect(capturedPermissionRequest).not.toBeNull();
+      const optionIds = capturedPermissionRequest.options.map((o: any) => o.optionId);
+      expect(optionIds).toContain("auto");
+    });
+  });
 });
--- a/src/tests/session-load.test.ts
+++ b/src/tests/session-load.test.ts
@@ -0,0 +1,333 @@
+import { describe, it, expect } from "vitest";
+import { randomUUID } from "crypto";
+import { AgentSideConnection, RequestError, SessionNotification } from "@agentclientprotocol/sdk";
+import { query, getSessionMessages } from "@anthropic-ai/claude-agent-sdk";
+import { ClaudeAcpAgent } from "../acp-agent.js";
+import { Pushable } from "../utils.js";
+import type { SDKUserMessage } from "@anthropic-ai/claude-agent-sdk";
+
+function createMockClient(): AgentSideConnection {
+  return {
+    sessionUpdate: async (_notification: SessionNotification) => {},
+    requestPermission: async () => ({ outcome: { outcome: "cancelled" } }),
+    readTextFile: async () => ({ content: "" }),
+    writeTextFile: async () => ({}),
+  } as unknown as AgentSideConnection;
+}
+
+describe.skipIf(!process.env.RUN_INTEGRATION_TESTS)("session load/resume lifecycle", () => {
+  it("SDK: session created but never prompted has no messages and is not resumable", async () => {
+    // Create a session via the SDK, initialize it, but never send a prompt
+    const sessionId = randomUUID();
+    const input = new Pushable<SDKUserMessage>();
+
+    const q = query({
+      prompt: input,
+      options: {
+        systemPrompt: { type: "preset", preset: "claude_code" },
+        sessionId,
+        settingSources: ["user", "project", "local"],
+        includePartialMessages: true,
+      },
+    });
+
+    // initializationResult() works without needing a prompt pushed
+    const initResult = await q.initializationResult();
+    expect(initResult).toBeDefined();
+
+    // Close without ever prompting
+    input.end();
+    q.return(undefined);
+
+    // Verify no messages were stored
+    const messages = await getSessionMessages(sessionId);
+    expect(messages).toEqual([]);
+
+    // Verify the session is not resumable
+    const input2 = new Pushable<SDKUserMessage>();
+    const q2 = query({
+      prompt: input2,
+      options: {
+        systemPrompt: { type: "preset", preset: "claude_code" },
+        resume: sessionId,
+        settingSources: ["user", "project", "local"],
+        includePartialMessages: true,
+      },
+    });
+
+    await expect(q2.initializationResult()).rejects.toThrow(
+      /No conversation found with session ID/,
+    );
+
+    input2.end();
+    q2.return(undefined);
+  }, 30000);
+
+  it("ACP: loadSession throws resourceNotFound for a non-existent session", async () => {
+    const agent = new ClaudeAcpAgent(createMockClient());
+    const bogusSessionId = randomUUID();
+
+    try {
+      await expect(
+        agent.loadSession({
+          sessionId: bogusSessionId,
+          cwd: process.cwd(),
+          mcpServers: [],
+        }),
+      ).rejects.toThrow(RequestError);
+    } finally {
+      await agent.dispose();
+    }
+  }, 30000);
+
+  it("ACP: resumeSession throws resourceNotFound for a non-existent session", async () => {
+    const agent = new ClaudeAcpAgent(createMockClient());
+    const bogusSessionId = randomUUID();
+
+    try {
+      await expect(
+        agent.resumeSession({
+          sessionId: bogusSessionId,
+          cwd: process.cwd(),
+          mcpServers: [],
+        }),
+      ).rejects.toThrow(RequestError);
+    } finally {
+      await agent.dispose();
+    }
+  }, 30000);
+
+  it("ACP: newSession without prompt, then loadSession on fresh agent throws resourceNotFound", async () => {
+    // Step 1: Create a real session via ACP, never prompt, dispose
+    const agentA = new ClaudeAcpAgent(createMockClient());
+    const { sessionId } = await agentA.newSession({
+      cwd: process.cwd(),
+      mcpServers: [],
+    });
+    expect(sessionId).toBeDefined();
+    await agentA.dispose();
+
+    // Step 2: Fresh agent tries to load that session
+    const agentB = new ClaudeAcpAgent(createMockClient());
+
+    try {
+      await expect(
+        agentB.loadSession({
+          sessionId,
+          cwd: process.cwd(),
+          mcpServers: [],
+        }),
+      ).rejects.toThrow(RequestError);
+    } finally {
+      await agentB.dispose();
+    }
+  }, 30000);
+
+  it("ACP: newSession without prompt, then resumeSession on fresh agent throws resourceNotFound", async () => {
+    const agentA = new ClaudeAcpAgent(createMockClient());
+    const { sessionId } = await agentA.newSession({
+      cwd: process.cwd(),
+      mcpServers: [],
+    });
+    await agentA.dispose();
+
+    const agentB = new ClaudeAcpAgent(createMockClient());
+
+    try {
+      await expect(
+        agentB.resumeSession({
+          sessionId,
+          cwd: process.cwd(),
+          mcpServers: [],
+        }),
+      ).rejects.toThrow(RequestError);
+    } finally {
+      await agentB.dispose();
+    }
+  }, 30000);
+
+  // Regression test for https://github.com/zed-industries/claude-code-acp/issues/579
+  // The client (Zed) renders its own local slash commands — e.g. `/model` —
+  // by injecting user-message prompts whose text is wrapped in
+  // `<command-name>` / `<local-command-stdout>` markers. The Claude SDK
+  // persists those messages verbatim in the session transcript. Without
+  // filtering, `loadSession` replays the markers as user_message_chunks
+  // ahead of the real prompt, leaking CLI internals into the transcript.
+  it("ACP: loadSession does not replay local-command metadata user messages", async () => {
+    const recordedUserChunks: string[] = [];
+    const client = {
+      sessionUpdate: async (notification: SessionNotification) => {
+        if (notification.update.sessionUpdate === "user_message_chunk") {
+          const content = notification.update.content;
+          if (content.type === "text") recordedUserChunks.push(content.text);
+        }
+      },
+      requestPermission: async () => ({ outcome: { outcome: "cancelled" } }),
+      readTextFile: async () => ({ content: "" }),
+      writeTextFile: async () => ({}),
+    } as unknown as AgentSideConnection;
+
+    const commandName =
+      "<command-name>/model</command-name>\n            <command-message>model</command-message>\n            <command-args>opus</command-args>";
+    const commandStdout =
+      "<local-command-stdout>Set model to opus (claude-opus-4-7)</local-command-stdout>";
+
+    // Step 1: create a session via the SDK, push the marker user messages
+    // followed by a real "hi" prompt. We bypass the ACP agent's prompt loop
+    // so we're only testing what the loadSession replay path does, not how
+    // prompts get routed.
+    const sessionId = randomUUID();
+    const input = new Pushable<SDKUserMessage>();
+    const q = query({
+      prompt: input,
+      options: {
+        systemPrompt: { type: "preset", preset: "claude_code" },
+        sessionId,
+        settingSources: ["user", "project", "local"],
+        includePartialMessages: true,
+      },
+    });
+    await q.initializationResult();
+
+    // Zed renders its local slash commands by mixing the marker blocks into
+    // the user's prompt content array alongside the real text. The Claude
+    // SDK persists the message verbatim — including the marker blocks —
+    // which is what loadSession must filter when it replays.
+    input.push({
+      type: "user",
+      message: {
+        role: "user",
+        content: [
+          { type: "text", text: commandName },
+          { type: "text", text: commandStdout },
+          { type: "text", text: "hi" },
+        ],
+      },
+      session_id: sessionId,
+      parent_tool_use_id: null,
+    });
+
+    for await (const msg of q) {
+      if (msg.type === "result") break;
+    }
+    input.end();
+    q.return(undefined);
+
+    // Sanity check: the SDK transcript contains the metadata we're filtering.
+    const sdkMessages = await getSessionMessages(sessionId);
+    const sdkTexts = sdkMessages
+      .map((m) => {
+        const c = (m as { message?: { content?: unknown } }).message?.content;
+        if (typeof c === "string") return c;
+        if (Array.isArray(c))
+          return c
+            .map((b) =>
+              b && typeof b === "object" && "text" in b
+                ? String((b as { text: unknown }).text)
+                : "",
+            )
+            .join("");
+        return "";
+      })
+      .join("\n");
+    expect(sdkTexts).toContain("<command-name>");
+    expect(sdkTexts).toContain("<local-command-stdout>");
+    expect(sdkTexts).toContain("hi");
+
+    // Step 2: load the session through the ACP agent and confirm the markers
+    // never reach the client as user_message_chunks, while the real "hi"
+    // prompt does.
+    const agent = new ClaudeAcpAgent(client);
+    try {
+      await agent.loadSession({
+        sessionId,
+        cwd: process.cwd(),
+        mcpServers: [],
+      });
+    } finally {
+      await agent.dispose();
+    }
+
+    for (const chunk of recordedUserChunks) {
+      expect(chunk).not.toContain("<command-name>");
+      expect(chunk).not.toContain("<command-message>");
+      expect(chunk).not.toContain("<command-args>");
+      expect(chunk).not.toContain("<local-command-stdout>");
+      expect(chunk).not.toContain("<local-command-stderr>");
+    }
+    expect(recordedUserChunks.some((c) => c.includes("hi"))).toBe(true);
+  }, 60000);
+
+  // Second regression: the original issue showed the markers and the real
+  // "hi" prompt all concatenated into a single string ending with "hi".
+  // loadSession must strip the marker tags in-place rather than dropping the
+  // whole message.
+  it("ACP: loadSession strips marker tags from a concatenated single-string prompt", async () => {
+    const recordedUserChunks: string[] = [];
+    const client = {
+      sessionUpdate: async (notification: SessionNotification) => {
+        if (notification.update.sessionUpdate === "user_message_chunk") {
+          const content = notification.update.content;
+          if (content.type === "text") recordedUserChunks.push(content.text);
+        }
+      },
+      requestPermission: async () => ({ outcome: { outcome: "cancelled" } }),
+      readTextFile: async () => ({ content: "" }),
+      writeTextFile: async () => ({}),
+    } as unknown as AgentSideConnection;
+
+    // Exact shape from https://github.com/zed-industries/claude-code-acp/issues/579.
+    const concatenated =
+      "<command-name>/model</command-name>\n            <command-message>model</command-message>\n            <command-args>opus</command-args>" +
+      "<local-command-stdout>Set model to opus (claude-opus-4-7)</local-command-stdout>\n" +
+      "<command-name>/model</command-name>\n            <command-message>model</command-message>\n            <command-args>opus[1m]</command-args>" +
+      "<local-command-stdout>Set model to opus[1m] (claude-opus-4-7[1m])</local-command-stdout>" +
+      "hi";
+
+    const sessionId = randomUUID();
+    const input = new Pushable<SDKUserMessage>();
+    const q = query({
+      prompt: input,
+      options: {
+        systemPrompt: { type: "preset", preset: "claude_code" },
+        sessionId,
+        settingSources: ["user", "project", "local"],
+        includePartialMessages: true,
+      },
+    });
+    await q.initializationResult();
+
+    input.push({
+      type: "user",
+      message: { role: "user", content: concatenated },
+      session_id: sessionId,
+      parent_tool_use_id: null,
+    });
+    for await (const msg of q) {
+      if (msg.type === "result") break;
+    }
+    input.end();
+    q.return(undefined);
+
+    const agent = new ClaudeAcpAgent(client);
+    try {
+      await agent.loadSession({
+        sessionId,
+        cwd: process.cwd(),
+        mcpServers: [],
+      });
+    } finally {
+      await agent.dispose();
+    }
+
+    expect(recordedUserChunks.length).toBeGreaterThan(0);
+    for (const chunk of recordedUserChunks) {
+      expect(chunk).not.toContain("<command-name>");
+      expect(chunk).not.toContain("<command-message>");
+      expect(chunk).not.toContain("<command-args>");
+      expect(chunk).not.toContain("<local-command-stdout>");
+      expect(chunk).not.toContain("<local-command-stderr>");
+    }
+    expect(recordedUserChunks.some((c) => c.includes("hi"))).toBe(true);
+  }, 60000);
+});
--- a/src/tests/tools.test.ts
+++ b/src/tests/tools.test.ts
@@ -11,7 +11,12 @@ import {
  BetaBashCodeExecutionToolResultBlockParam,
 } from "@anthropic-ai/sdk/resources/beta.mjs";
 import { toAcpNotifications, ToolUseCache, Logger } from "../acp-agent.js";
-import { toolUpdateFromToolResult, createPostToolUseHook } from "../tools.js";
+import {
+  toolUpdateFromToolResult,
+  createPostToolUseHook,
+  toolInfoFromToolUse,
+  planEntries,
+} from "../tools.js";

 describe("rawOutput in tool call updates", () => {
  const mockClient = {} as AgentSideConnection;
@@ -1051,6 +1056,162 @@ describe("Bash terminal output", () => {
    });
  });

+  describe("post-tool-use hook sends diff content for Write tool", () => {
+    // Regression: previously the dispatch only invoked
+    // toolUpdateFromDiffToolResponse for `Edit`, so a Write that
+    // overwrote an existing file (FileWriteOutput.type === "update")
+    // showed a "creation" diff (oldText: null, full new content) at
+    // tool_use time and was never corrected after the tool ran.
+    it("should emit a real diff for Write of an existing file (type: update)", async () => {
+      const toolUseCache: ToolUseCache = {};
+      const hookUpdates: any[] = [];
+      const mockClientWithUpdate = {
+        sessionUpdate: async (notification: any) => {
+          hookUpdates.push(notification);
+        },
+      } as unknown as AgentSideConnection;
+
+      toAcpNotifications(
+        [
+          {
+            type: "tool_use" as const,
+            id: "toolu_write_update",
+            name: "Write",
+            input: {
+              file_path: "/Users/test/project/file.ts",
+              content: "line1\nNEW line2\nline3",
+            },
+          },
+        ],
+        "assistant",
+        "test-session",
+        toolUseCache,
+        mockClientWithUpdate,
+        mockLogger,
+      );
+
+      const hook = createPostToolUseHook(mockLogger);
+      await hook(
+        {
+          hook_event_name: "PostToolUse",
+          tool_name: "Write",
+          tool_input: {
+            file_path: "/Users/test/project/file.ts",
+            content: "line1\nNEW line2\nline3",
+          },
+          tool_response: {
+            type: "update",
+            filePath: "/Users/test/project/file.ts",
+            content: "line1\nNEW line2\nline3",
+            originalFile: "line1\nold line2\nline3",
+            structuredPatch: [
+              {
+                oldStart: 1,
+                oldLines: 3,
+                newStart: 1,
+                newLines: 3,
+                lines: [" line1", "-old line2", "+NEW line2", " line3"],
+              },
+            ],
+            userModified: false,
+          },
+          tool_use_id: "toolu_write_update",
+          session_id: "test-session",
+          transcript_path: "/tmp/test",
+          cwd: "/tmp",
+        },
+        "toolu_write_update",
+        { signal: AbortSignal.abort() },
+      );
+
+      expect(hookUpdates).toHaveLength(1);
+      const hookUpdate = hookUpdates[0].update;
+      expect(hookUpdate._meta.claudeCode.toolName).toBe("Write");
+      expect(hookUpdate.content).toEqual([
+        {
+          type: "diff",
+          path: "/Users/test/project/file.ts",
+          oldText: "line1\nold line2\nline3",
+          newText: "line1\nNEW line2\nline3",
+        },
+      ]);
+      expect(hookUpdate.locations).toEqual([{ path: "/Users/test/project/file.ts", line: 1 }]);
+    });
+
+    it("should still emit a sensible diff for Write of a brand-new file (type: create)", async () => {
+      const toolUseCache: ToolUseCache = {};
+      const hookUpdates: any[] = [];
+      const mockClientWithUpdate = {
+        sessionUpdate: async (notification: any) => {
+          hookUpdates.push(notification);
+        },
+      } as unknown as AgentSideConnection;
+
+      toAcpNotifications(
+        [
+          {
+            type: "tool_use" as const,
+            id: "toolu_write_create",
+            name: "Write",
+            input: {
+              file_path: "/Users/test/project/new.ts",
+              content: "first\nsecond",
+            },
+          },
+        ],
+        "assistant",
+        "test-session",
+        toolUseCache,
+        mockClientWithUpdate,
+        mockLogger,
+      );
+
+      const hook = createPostToolUseHook(mockLogger);
+      await hook(
+        {
+          hook_event_name: "PostToolUse",
+          tool_name: "Write",
+          tool_input: {
+            file_path: "/Users/test/project/new.ts",
+            content: "first\nsecond",
+          },
+          tool_response: {
+            type: "create",
+            filePath: "/Users/test/project/new.ts",
+            content: "first\nsecond",
+            originalFile: null,
+            structuredPatch: [
+              {
+                oldStart: 0,
+                oldLines: 0,
+                newStart: 1,
+                newLines: 2,
+                lines: ["+first", "+second"],
+              },
+            ],
+          },
+          tool_use_id: "toolu_write_create",
+          session_id: "test-session",
+          transcript_path: "/tmp/test",
+          cwd: "/tmp",
+        },
+        "toolu_write_create",
+        { signal: AbortSignal.abort() },
+      );
+
+      expect(hookUpdates).toHaveLength(1);
+      const hookUpdate = hookUpdates[0].update;
+      expect(hookUpdate.content).toEqual([
+        {
+          type: "diff",
+          path: "/Users/test/project/new.ts",
+          oldText: null,
+          newText: "first\nsecond",
+        },
+      ]);
+    });
+  });
+
  describe("post-tool-use hook preserves terminal _meta", () => {
    it("should send terminal_output and terminal_exit as separate notifications, and hook should only have claudeCode", async () => {
      const clientCapabilities: ClientCapabilities = {
@@ -1235,3 +1396,148 @@ describe("Bash terminal output", () => {
    });
  });
 });
+
+describe("toolInfoFromToolUse - ExitPlanMode", () => {
+  it("should include plan text in content when input.plan is provided", () => {
+    const toolUse = {
+      name: "ExitPlanMode",
+      id: "toolu_plan_1",
+      input: {
+        plan: "# My Plan\n\n## Step 1\nDo something",
+        planFilePath: "/tmp/plan.md",
+      },
+    };
+
+    const info = toolInfoFromToolUse(toolUse, false);
+
+    expect(info.kind).toBe("switch_mode");
+    expect(info.title).toBe("Ready to code?");
+    expect(info.content).toHaveLength(1);
+    expect(info.content![0]).toEqual({
+      type: "content",
+      content: { type: "text", text: "# My Plan\n\n## Step 1\nDo something" },
+    });
+  });
+
+  it("should return empty content when input.plan is not provided", () => {
+    const toolUse = {
+      name: "ExitPlanMode",
+      id: "toolu_plan_2",
+      input: {},
+    };
+
+    const info = toolInfoFromToolUse(toolUse, false);
+
+    expect(info.kind).toBe("switch_mode");
+    expect(info.content).toEqual([]);
+  });
+});
+
+describe("toolInfoFromToolUse - undefined input regression", () => {
+  it("Read with undefined input should not throw", () => {
+    const toolUse = { name: "Read", id: "toolu_read_undef", input: undefined };
+    const info = toolInfoFromToolUse(toolUse, false);
+    expect(info.title).toBe("Read File");
+    expect(info.locations).toEqual([]);
+  });
+
+  it("Grep with undefined input should not throw", () => {
+    const toolUse = { name: "Grep", id: "toolu_grep_undef", input: undefined };
+    const info = toolInfoFromToolUse(toolUse, false);
+    expect(info.title).toBe("grep");
+  });
+
+  it("Glob with undefined input should not throw", () => {
+    const toolUse = { name: "Glob", id: "toolu_glob_undef", input: undefined };
+    const info = toolInfoFromToolUse(toolUse, false);
+    expect(info.title).toBe("Find");
+    expect(info.locations).toEqual([]);
+  });
+
+  it("WebSearch with undefined input should not throw", () => {
+    const toolUse = { name: "WebSearch", id: "toolu_ws_undef", input: undefined };
+    const info = toolInfoFromToolUse(toolUse, false);
+    expect(info.title).toBe("Web search");
+  });
+
+  it("TodoWrite with undefined input should not throw", () => {
+    const toolUse = { name: "TodoWrite", id: "toolu_todo_undef", input: undefined };
+    const info = toolInfoFromToolUse(toolUse, false);
+    expect(info.title).toBe("Update TODOs");
+  });
+});
+
+describe("planEntries - undefined input regression", () => {
+  it("should return empty array when input is undefined", () => {
+    expect(planEntries(undefined)).toEqual([]);
+  });
+
+  it("should return empty array when input has no todos", () => {
+    expect(planEntries({} as any)).toEqual([]);
+  });
+
+  it("should still map valid todos correctly", () => {
+    const result = planEntries({
+      todos: [
+        { content: "Task 1", status: "pending", activeForm: "" },
+        { content: "Task 2", status: "completed", activeForm: "" },
+      ],
+    });
+    expect(result).toEqual([
+      { content: "Task 1", status: "pending", priority: "medium" },
+      { content: "Task 2", status: "completed", priority: "medium" },
+    ]);
+  });
+});
+
+describe("toAcpNotifications - TodoWrite with undefined input regression", () => {
+  const mockClient = {} as AgentSideConnection;
+  const mockLogger: Logger = { log: () => {}, error: () => {} };
+
+  it("should not throw when TodoWrite tool_use has undefined input", () => {
+    const toolUseCache: ToolUseCache = {};
+
+    const notifications = toAcpNotifications(
+      [
+        {
+          type: "tool_use" as const,
+          id: "toolu_todo_undef",
+          name: "TodoWrite",
+          input: undefined as any,
+        },
+      ],
+      "assistant",
+      "test-session",
+      toolUseCache,
+      mockClient,
+      mockLogger,
+    );
+
+    // TodoWrite with undefined input should not crash, and should not emit plan update
+    const planUpdates = notifications.filter((n) => (n.update as any).sessionUpdate === "plan");
+    expect(planUpdates).toHaveLength(0);
+  });
+
+  it("should still emit plan update when TodoWrite has valid input", () => {
+    const toolUseCache: ToolUseCache = {};
+
+    const notifications = toAcpNotifications(
+      [
+        {
+          type: "tool_use" as const,
+          id: "toolu_todo_valid",
+          name: "TodoWrite",
+          input: { todos: [{ content: "Do X", status: "pending" }] },
+        },
+      ],
+      "assistant",
+      "test-session",
+      toolUseCache,
+      mockClient,
+      mockLogger,
+    );
+
+    const planUpdates = notifications.filter((n) => (n.update as any).sessionUpdate === "plan");
+    expect(planUpdates).toHaveLength(1);
+  });
+});
--- a/src/tools.ts
+++ b/src/tools.ts
@@ -5,6 +5,19 @@ import {
  ToolCallLocation,
  ToolKind,
 } from "@agentclientprotocol/sdk";
+import { HookCallback } from "@anthropic-ai/claude-agent-sdk";
+import {
+  AgentInput,
+  BashInput,
+  FileEditInput,
+  FileReadInput,
+  FileWriteInput,
+  GlobInput,
+  GrepInput,
+  TodoWriteInput,
+  WebFetchInput,
+  WebSearchInput,
+} from "@anthropic-ai/claude-agent-sdk/sdk-tools.js";
 import {
  ImageBlockParam,
  TextBlockParam,
@@ -14,29 +27,31 @@ import {
  WebSearchToolResultError,
 } from "@anthropic-ai/sdk/resources";
 import {
-  BetaBashCodeExecutionToolResultBlockParam,
  BetaBashCodeExecutionResultBlock,
+  BetaBashCodeExecutionToolResultBlockParam,
  BetaBashCodeExecutionToolResultError,
-  BetaCodeExecutionToolResultBlockParam,
  BetaCodeExecutionResultBlock,
+  BetaCodeExecutionToolResultBlockParam,
  BetaCodeExecutionToolResultError,
+  BetaImageBlockParam,
  BetaRequestMCPToolResultBlockParam,
-  BetaTextEditorCodeExecutionToolResultBlockParam,
-  BetaTextEditorCodeExecutionViewResultBlock,
  BetaTextEditorCodeExecutionCreateResultBlock,
  BetaTextEditorCodeExecutionStrReplaceResultBlock,
+  BetaTextEditorCodeExecutionToolResultBlockParam,
  BetaTextEditorCodeExecutionToolResultError,
+  BetaTextEditorCodeExecutionViewResultBlock,
+  BetaToolReferenceBlock,
  BetaToolResultBlockParam,
  BetaToolSearchToolResultBlockParam,
-  BetaToolReferenceBlock,
-  BetaToolSearchToolSearchResultBlock,
  BetaToolSearchToolResultError,
-  BetaWebFetchToolResultBlockParam,
+  BetaToolSearchToolSearchResultBlock,
  BetaWebFetchBlock,
+  BetaWebFetchToolResultBlockParam,
  BetaWebFetchToolResultErrorBlock,
  BetaWebSearchToolResultBlockParam,
-  BetaImageBlockParam,
 } from "@anthropic-ai/sdk/resources/beta.mjs";
+import path from "node:path";
+import { Logger } from "./acp-agent.js";

 /**
 * Union of all possible content types that can appear in tool results from the Anthropic SDK.
@@ -61,20 +76,6 @@ type ToolResultContent =
  | BetaTextEditorCodeExecutionCreateResultBlock
  | BetaTextEditorCodeExecutionStrReplaceResultBlock
  | BetaTextEditorCodeExecutionToolResultError;
-import { HookCallback } from "@anthropic-ai/claude-agent-sdk";
-import { Logger } from "./acp-agent.js";
-import {
-  AgentInput,
-  BashInput,
-  FileEditInput,
-  FileReadInput,
-  FileWriteInput,
-  GlobInput,
-  GrepInput,
-  TodoWriteInput,
-  WebFetchInput,
-  WebSearchInput,
-} from "@anthropic-ai/claude-agent-sdk/sdk-tools.js";

 interface ToolInfo {
  title: string;
@@ -103,15 +104,31 @@ interface ToolUpdate {
  };
 }

+/**
+ * Convert an absolute file path to a project-relative path for display.
+ * Returns the original path if it's outside the project directory or if no cwd is provided.
+ */
+export function toDisplayPath(filePath: string, cwd?: string): string {
+  if (!cwd) return filePath;
+  const resolvedCwd = path.resolve(cwd);
+  const resolvedFile = path.resolve(filePath);
+  if (resolvedFile.startsWith(resolvedCwd + path.sep) || resolvedFile === resolvedCwd) {
+    return path.relative(resolvedCwd, resolvedFile);
+  }
+  return filePath;
+}
+
 export function toolInfoFromToolUse(
  toolUse: any,
  supportsTerminalOutput: boolean = false,
+  cwd?: string,
 ): ToolInfo {
  const name = toolUse.name;

  switch (name) {
+    case "Agent":
    case "Task": {
-      const input = toolUse.input as AgentInput | BashInput;
+      const input = toolUse.input as AgentInput | BashInput | undefined;
      return {
        title: input?.description ? input.description : "Task",
        kind: "think",
@@ -128,7 +145,7 @@ export function toolInfoFromToolUse(
    }

    case "Bash": {
-      const input = toolUse.input as BashInput;
+      const input = toolUse.input as BashInput | undefined;
      return {
        title: input?.command ? input.command : "Terminal",
        kind: "execute",
@@ -146,17 +163,18 @@ export function toolInfoFromToolUse(
    }

    case "Read": {
-      const input = toolUse.input as FileReadInput;
+      const input = toolUse.input as FileReadInput | undefined;
      let limit = "";
-      if (input.limit && input.limit > 0) {
+      if (input?.limit && input.limit > 0) {
        limit = " (" + (input.offset ?? 1) + " - " + ((input.offset ?? 1) + input.limit - 1) + ")";
-      } else if (input.offset) {
+      } else if (input?.offset) {
        limit = " (from line " + input.offset + ")";
      }
+      const displayPath = input?.file_path ? toDisplayPath(input.file_path, cwd) : "File";
      return {
-        title: "Read " + (input.file_path ?? "File") + limit,
+        title: "Read " + displayPath + limit,
        kind: "read",
-        locations: input.file_path
+        locations: input?.file_path
          ? [
              {
                path: input.file_path,
@@ -169,7 +187,7 @@ export function toolInfoFromToolUse(
    }

    case "Write": {
-      const input = toolUse.input as FileWriteInput;
+      const input = toolUse.input as FileWriteInput | undefined;
      let content: ToolCallContent[] = [];
      if (input && input.file_path) {
        content = [
@@ -188,8 +206,9 @@ export function toolInfoFromToolUse(
          },
        ];
      }
+      const displayPath = input?.file_path ? toDisplayPath(input.file_path, cwd) : undefined;
      return {
-        title: input?.file_path ? `Write ${input.file_path}` : "Write",
+        title: displayPath ? `Write ${displayPath}` : "Write",
        kind: "edit",
        content,
        locations: input?.file_path ? [{ path: input.file_path }] : [],
@@ -197,7 +216,7 @@ export function toolInfoFromToolUse(
    }

    case "Edit": {
-      const input = toolUse.input as FileEditInput;
+      const input = toolUse.input as FileEditInput | undefined;
      let content: ToolCallContent[] = [];
      if (input && input.file_path && (input.old_string || input.new_string)) {
        content = [
@@ -209,8 +228,9 @@ export function toolInfoFromToolUse(
          },
        ];
      }
+      const displayPath = input?.file_path ? toDisplayPath(input.file_path, cwd) : undefined;
      return {
-        title: input?.file_path ? `Edit ${input.file_path}` : "Edit",
+        title: displayPath ? `Edit ${displayPath}` : "Edit",
        kind: "edit",
        content,
        locations: input?.file_path ? [{ path: input.file_path }] : [],
@@ -218,44 +238,44 @@ export function toolInfoFromToolUse(
    }

    case "Glob": {
-      const input = toolUse.input as GlobInput;
+      const input = toolUse.input as GlobInput | undefined;
      let label = "Find";
-      if (input.path) {
+      if (input?.path) {
        label += ` \`${input.path}\``;
      }
-      if (input.pattern) {
+      if (input?.pattern) {
        label += ` \`${input.pattern}\``;
      }
      return {
        title: label,
        kind: "search",
        content: [],
-        locations: input.path ? [{ path: input.path }] : [],
+        locations: input?.path ? [{ path: input.path }] : [],
      };
    }

    case "Grep": {
-      const input = toolUse.input as GrepInput;
+      const input = toolUse.input as GrepInput | undefined;
      let label = "grep";

-      if (input["-i"]) {
+      if (input?.["-i"]) {
        label += " -i";
      }
-      if (input["-n"]) {
+      if (input?.["-n"]) {
        label += " -n";
      }

-      if (input["-A"] !== undefined) {
+      if (input?.["-A"] !== undefined) {
        label += ` -A ${input["-A"]}`;
      }
-      if (input["-B"] !== undefined) {
+      if (input?.["-B"] !== undefined) {
        label += ` -B ${input["-B"]}`;
      }
-      if (input["-C"] !== undefined) {
+      if (input?.["-C"] !== undefined) {
        label += ` -C ${input["-C"]}`;
      }

-      if (input.output_mode) {
+      if (input?.output_mode) {
        switch (input.output_mode) {
          case "files_with_matches":
            label += " -l";
@@ -269,27 +289,27 @@ export function toolInfoFromToolUse(
        }
      }

-      if (input.head_limit !== undefined) {
+      if (input?.head_limit !== undefined) {
        label += ` | head -${input.head_limit}`;
      }

-      if (input.glob) {
+      if (input?.glob) {
        label += ` --include="${input.glob}"`;
      }

-      if (input.type) {
+      if (input?.type) {
        label += ` --type=${input.type}`;
      }

-      if (input.multiline) {
+      if (input?.multiline) {
        label += " -P";
      }

-      if (input.pattern) {
+      if (input?.pattern) {
        label += ` "${input.pattern}"`;
      }

-      if (input.path) {
+      if (input?.path) {
        label += ` ${input.path}`;
      }

@@ -318,14 +338,14 @@ export function toolInfoFromToolUse(
    }

    case "WebSearch": {
-      const input = toolUse.input as WebSearchInput;
-      let label = `"${input.query}"`;
+      const input = toolUse.input as WebSearchInput | undefined;
+      let label = input?.query ? `"${input.query}"` : "Web search";

-      if (input.allowed_domains && input.allowed_domains.length > 0) {
+      if (input?.allowed_domains && input.allowed_domains.length > 0) {
        label += ` (allowed: ${input.allowed_domains.join(", ")})`;
      }

-      if (input.blocked_domains && input.blocked_domains.length > 0) {
+      if (input?.blocked_domains && input.blocked_domains.length > 0) {
        label += ` (blocked: ${input.blocked_domains.join(", ")})`;
      }

@@ -337,7 +357,7 @@ export function toolInfoFromToolUse(
    }

    case "TodoWrite": {
-      const input = toolUse.input as TodoWriteInput;
+      const input = toolUse.input as TodoWriteInput | undefined;
      return {
        title: Array.isArray(input?.todos)
          ? `Update TODOs: ${input.todos.map((todo: any) => todo.content).join(", ")}`
@@ -348,10 +368,13 @@ export function toolInfoFromToolUse(
    }

    case "ExitPlanMode": {
+      const planInput = toolUse.input as { plan?: string } | undefined;
      return {
        title: "Ready to code?",
        kind: "switch_mode",
-        content: [],
+        content: planInput?.plan
+          ? [{ type: "content" as const, content: { type: "text" as const, text: planInput.plan } }]
+          : [],
      };
    }

@@ -461,7 +484,7 @@ export function toolUpdateFromToolResult(
        result.type === "bash_code_execution_result"
      ) {
        const bashResult = result as BetaBashCodeExecutionResultBlock;
-        output = bashResult.stdout || bashResult.stderr || "";
+        output = [bashResult.stdout, bashResult.stderr].filter(Boolean).join("\n");
        exitCode = bashResult.return_code;
      } else if (typeof result === "string") {
        output = result;
@@ -638,10 +661,10 @@ export type ClaudePlanEntry = {
  activeForm: string;
 };

-export function planEntries(input: { todos: ClaudePlanEntry[] }): PlanEntry[] {
-  return input.todos.map((input) => ({
-    content: input.content,
-    status: input.status,
+export function planEntries(input: { todos: ClaudePlanEntry[] } | undefined): PlanEntry[] {
+  return (input?.todos ?? []).map((todo) => ({
+    content: todo.content,
+    status: todo.status,
    priority: "medium",
  }));
 }
@@ -656,7 +679,7 @@ export function markdownEscape(text: string): string {
  return escape + "\n" + text + (text.endsWith("\n") ? "" : "\n") + escape;
 }

-interface EditToolResponseHunk {
+interface DiffToolResponseHunk {
  oldStart: number;
  oldLines: number;
  newStart: number;
@@ -664,23 +687,24 @@ interface EditToolResponseHunk {
  lines: string[];
 }

-interface EditToolResponse {
+interface DiffToolResponse {
  filePath?: string;
-  structuredPatch?: EditToolResponseHunk[];
+  structuredPatch?: DiffToolResponseHunk[];
 }

 /**
- * Builds diff ToolUpdate content from the structured Edit toolResponse provided
- * by the PostToolUse hook. Unlike parsing the plain unified diff string, this uses
- * the pre-parsed structuredPatch which supports multiple replacement sites (replaceAll)
- * and always includes context lines for better readability.
+ * Builds diff ToolUpdate content from the structured toolResponse provided by
+ * the PostToolUse hook for diff-producing tools (Edit, Write). Unlike parsing
+ * the plain unified diff string, this uses the pre-parsed structuredPatch
+ * which supports multiple replacement sites (replaceAll) and always includes
+ * context lines for better readability.
 */
-export function toolUpdateFromEditToolResponse(toolResponse: unknown): {
+export function toolUpdateFromDiffToolResponse(toolResponse: unknown): {
  content?: ToolCallContent[];
  locations?: ToolCallLocation[];
 } {
  if (!toolResponse || typeof toolResponse !== "object") return {};
-  const response = toolResponse as EditToolResponse;
+  const response = toolResponse as DiffToolResponse;
  if (!response.filePath || !Array.isArray(response.structuredPatch)) return {};

  const content: ToolCallContent[] = [];