## Summary
In an ACP client, `/mcp` showed the CLI text "… Use `/mcp` in the
terminal for details." The user of an ACP client has no terminal. In SDK
mode, Claude Code also refuses `/mcp reconnect|enable|disable` with
"Reconnect, enable, and disable aren't available in this session." and
changes nothing.
- `/mcp` and `/mcp reconnect|enable|disable [<server>|all]` still run as
a normal Claude Code turn. The adapter replaces the text of the local
command, through the same hook that `/usage` uses. Prompt order, cancel,
and the provider update stay with the turn.
- For an action, the adapter runs it in that turn through the SDK
control API (`reconnectMcpServer`, `toggleMcpServer`). It shows one
result line for each server, then the status list.
- The status list comes from `query.mcpServerStatus()`. It has a summary
line and groups by status: failed, needs authentication, connecting,
connected, and disabled on one line. A failed server shows a short
cleaned error. A list is used and not a table, because a table breaks
words in narrow columns.
- Useful CLI sentences stay, for example "There's no MCP server named
…". Sentences that point to the terminal are removed.
- After an explicit reconnect, a server from the ACP session that needs
authentication starts the existing MCP OAuth flow, when the client
supports URL elicitation.
- `/mcp` is advertised with the input hint `[reconnect|enable|disable
[<server>|all]]` and is hidden on history replay.
## Other commits in this branch
This branch also carries commits of other work: skill command metadata
for AIR, a faster `session/load` replay, and the resumed permission
mode. They can move to separate PRs if preferred.
## Tests
- `src/tests/mcp-command.test.ts` (61 tests). The fake CLI answers with
the real SDK-mode texts and message shapes (synthetic assistant +
result) and does not change any status.
- A live run through the adapter with a broken and a working stdio
server: `/mcp`, `/mcp disable good`, `/mcp enable good`, `/mcp reconnect
broken`, `/mcp reconnect nosuch`.
- `compare.ts` allows one new difference against the origin/main
baseline: the `mcp` command in `available_commands_update`. The baseline
files are unchanged.
- `npm run check` and `npm run test:run` pass.
Released AIR versions, such as 262.air/release-09, read the custom
answer companion only from the root key
_meta._askUserQuestionCustomAnswer. The adapter sent only
_meta.jetbrains.air.customAnswer, so these versions showed the companion
as a separate question.
For an AIR client the companion now carries both keys with the same
value. Other clients get no key, as before. Remove the root key when the
ACP registry no longer serves AIR versions that read only that key.
IJAI-2892
## Summary
One PR with separate commits (the earlier stack #1168 and #1169 is
folded back in).
### Exact git patches
- `src/diff.ts` builds exact git patches for Edit and Write: real line
numbers, file modes, CRLF, the final-newline marker, a size limit
(including a bound for `replace_all`), binary detection, and an
asynchronous diff with a time budget. The approval preview mirrors the
input normalisation of the bundled CLI.
- No creation patch over an existing file or for an Edit of an existing
empty file, and no patch for a Write that changed nothing.
### For every client
Clients that do not declare `_meta.jetbrains.air` (Zed, plain ACP
clients) get the same information as on `main`.
- A `tool_call_update` omits top-level fields that did not change since
the last report of the same tool call (ACP merge semantics). The `_meta`
of each report stays complete.
- Bug fixes: a failed approval preview no longer blanks the approval
diff; a permission denial replaces a pinned approval patch; the
compaction summary is not repeated when the streamed text is identical;
streamed subagent text is not sent twice; async task fields are recorded
only after a successful send; late updates of a finished native subagent
are dropped; the updates of a fresh Claude context after a clear-context
restart reach the ACP session.
### For AIR only
- One fact in one field: input in `rawInput`, the file text of an edit
only in the diff (`diffPatch`), output once, no display copies of the
input (`rawInputRendering`), `_meta.claudeCode` reduced to `toolName`,
`parentToolUseId` and the `toolResponse` markers, per-key `_meta`
filtering.
- Plan file mode (`planFile`): ExitPlanMode carries only
`rawInput.planFilePath`; AIR follows the file that Claude writes.
- AIR-only keys move under `_meta.jetbrains.air.*` and are no longer
sent to other clients.
### Structure and docs
- `src/tool-calls/`: one reporter per tool, one `ClientCapabilities`,
one `AcpToolCallRenderer`, the field tracker and the changed-meta
filter.
- `docs/air-extensions.md` replaces the per-extension AIR docs.
## Tests
- `npm run build`, `npm run check`, `npm run test:run`: 1995 passed.
- Scenario tests for plain, Zed, and AIR clients: the SDK mock honours
the adapter's session id, every update must reach its ACP session, every
message is validated against the ACP JSON schema with all formats
checked, plain and Zed are compared with a recorded `main` baseline, AIR
keeps compact golden recordings.
## Summary
- use the canonical tool-call title for permission prompts, so shell
approvals show the exact command instead of the model-authored
description
- map `PowerShell` through the same shell presentation path as `Bash`
- complete the PowerShell terminal lifecycle with matching
`terminal_info`, `terminal_output`, and `terminal_exit` metadata,
including failed commands
- add focused regression coverage without changing the existing Bash
lifecycle tests
Fixes#1068
## PowerShell input and result contract
`PowerShell` is a distinct tool name, but its documented input fields
match `Bash`: the executable command is in `command`, while
`description` is an optional human-readable summary. The Claude Code
hooks reference documents the shared `command`, `description`,
`timeout`, and `run_in_background` fields, and the tools reference
likewise states that PowerShell hooks receive the command in
`tool_input.command`.
Anthropic closed
[claude-code#83647](https://github.com/anthropics/claude-code/issues/83647#issuecomment-5310033500)
after documenting this equivalence. Reusing the SDK's exported
`BashInput` shape for the common input fields is therefore intentional;
the SDK currently does not export a separate `PowerShellInput` type.
PowerShell results do not need a PowerShell-specific output assumption
here. The result mapper first shape-checks structured stdout/stderr and
otherwise falls back to the standard raw `tool_result` string or
content-block array. Routing PowerShell through that existing mapper
preserves this fallback while completing the adapter's capability-gated
terminal sequence:
1. initial `tool_call` with `terminal_info`
2. `tool_call_update` with `terminal_output`
3. final `tool_call_update` with `terminal_exit`
All three events use the original tool-use ID, so the client can
reconcile and close the terminal. Failed PowerShell results also bypass
the generic error renderer and finish the terminal with a non-zero
display exit code.
The permission presentation introduced in #1004 already handled `Bash`
and `PowerShell` together when choosing a description-based permission
title, but the canonical tool presentation and terminal result paths
only handled `Bash`. This PR completes both paths.
References:
- [Claude Code hooks: PowerShell
input](https://code.claude.com/docs/en/hooks#powershell)
- [Claude Code tools reference: PowerShell
tool](https://code.claude.com/docs/en/tools-reference#powershell-tool)
- [Anthropic schema
confirmation](https://github.com/anthropics/claude-code/issues/83647#issuecomment-5310033500)
## Verification
- `npm run check`
- `npm run build`
- `npm run test:run -- src/tests/tools.test.ts -t "PowerShell terminal
output"` (1 passed)
- full suite with an empty temporary Claude config:
`CLAUDE_CONFIG_DIR=<empty temporary directory> npm run test:run` (1070
passed, 27 skipped)
---------
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
Co-authored-by: Ben Brandt <benjamin.j.brandt@gmail.com>
Claude supplies structured patches for Edit and Write, but ACP diff
blocks previously carried only the reconstructed texts. AIR had to
compare those texts again to display added and removed counts.
Count patch operations in the existing conversion loop and publish
`_meta.jetbrains.air.diffStats` on each diff block. This uses the same
contract as https://github.com/agentclientprotocol/codex-acp/pull/501.
Each block has its own counts. No extra traversal, full file comparison,
or navigation calculation is needed.
Omit statistics when patches are absent or inconsistent, so clients
retain their existing fallback. Exclude EOF markers from the displayed
texts. Document the extension and cover per-block counts, Edit and Write
hooks, EOF changes, and fallback behavior.
Validation:
- `npm run build`
- `npm run check`
- `env -u ANTHROPIC_BASE_URL npm run test:run` (the local endpoint
override affects unrelated provider tests)
## Summary
Closes#1119. Supersedes #1120 by @amansk (their commits are included on
this branch; the fork does not allow maintainer edits, so review fixes
landed here instead).
Hosts that never use `bypassPermissions` can opt a session out by
sending `_meta.claudeCode.options.allowDangerouslySkipPermissions:
false` on `session/new` (and again on `session/load`). When omitted or
`true`, behavior is unchanged: bypass is available on non-root sessions.
## Design
Bypass availability is decided once per session in `createSession`:
```ts
const allowBypass =
ALLOW_BYPASS && sessionMeta?.claudeCode?.options?.allowDangerouslySkipPermissions !== false;
```
and threaded to the three places that must agree:
- the SDK `allowDangerouslySkipPermissions` flag (the actual
enforcement; the CLI refuses bypass without it)
- the spawn-time mode clamp in `resolvePermissionMode`, so a
`permissions.defaultMode` of `bypassPermissions` (or a mode carried
across a clear-context restart) becomes `default` before `query()`
rather than failing at spawn
- the session mode catalog in `SessionModeManager.initialize`, so the
client is never offered a mode the CLI would reject
## Changes from #1120
- Fold the host opt-out into `resolvePermissionMode` via an
`allowBypass` parameter instead of a second inline clamp; drop the
separate `resolveAllowDangerouslySkipPermissions` helper
- Make `allowBypass` a required `initialize` parameter and remove the
module-global `ALLOW_BYPASS` fallback from `session-mode.ts`
- Fix `tsc` errors in the new tests (`modes` is optional on
`NewSessionResponse`), which broke the CI build job
- Move the settings-driven clamp test onto the shared
`acp-agent-settings` fixture (isolated cwd, no double module reload,
passes as root)
- Document the opt-out in `docs/permission-extension.md`
## Testing
- `npm run check`, `npm run build` — pass
- `npm run test:run` — pass (the branch is based before #1123, so the
SDK version guard in `model-presentation.test.ts` only passes with the
pinned 0.3.257; it clears on rebase)
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Amandeep Khurana <amansk@gmail.com>
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
Clients that negotiate the AIR `recommendedValue` capability receive
concrete model and effort choices instead of ambiguous default rows. The
SDK's default model is mapped only to an exact selectable model
(including equivalent context suffix spellings); unavailable
recommendations retain the legacy default. Recommendations remain
independent of explicit selections.
The adapter applies the displayed effort to the SDK at startup and on
model switches, preserving supported user pins and re-reading per-model
settings for automatic choices. Permission modes and clients without the
capability retain their existing configuration behavior. Model and
effort configuration logic is extracted into dedicated modules.
Terse SDK model labels are enriched with concrete versions from model
metadata: for example, `Fable` becomes `Fable 5.1`, `Sonnet` becomes
`Sonnet 5`, and `Opus (1M context)` becomes `Opus 5`. Custom names are
preserved, and collisions fall back to the original labels. An SDK
version guard requires review on upgrades, and an opt-in live
initialization test verifies the upstream model contract without sending
a prompt.
Validation:
- `npm run check`
- `npm run build`
- `env -u ANTHROPIC_BASE_URL npm run test:run` — 1253 passed, 28 skipped
- `RUN_INTEGRATION_TESTS=true npx vitest run
src/tests/model-presentation.test.ts` — 4 passed
- Regression coverage for unavailable model generations/context windows,
startup effort, per-model settings, unsupported effort pins, and
duplicate Opus labels
Today nothing reaches npm between two release-please releases, so trying
a merged
but unreleased change means building from source. This adds a
continuously
published preview channel.
Every CI-green push to `main` that is not a release merge publishes
`<package.json version, patch bumped>-preview.N` under the `preview`
dist-tag,
tags the commit `v<version>`, and dispatches the same agent registry
update a
stable release does. With `main` at 0.73.0 that gives
`0.73.1-preview.1`,
`0.73.1-preview.2`, and so on; `N` restarts at 1 whenever a release
moves the
base. `latest` is untouched.
## Approach
- **Added to `publish.yml` rather than a new workflow.** npm binds a
trusted
publisher to one repo + workflow filename, and a package gets only one.
A
separate file would fail OIDC and cost the stable path its publisher.
- **Gated on the `CI` workflow succeeding** (`workflow_run`), so a red
commit is
never published. release-please's release merge is excluded by the head
commit's author and subject — its `releases_created` output is not
reachable
from a different workflow run.
- **`N` is the max of the npm registry and local git tags.** npm
reserves
versions forever, even after `unpublish`, so a reused `N` is a hard
failure;
but the registry is CDN-served and lags, which the tags cover. A
registry read
failure aborts rather than guessing.
- **Base is patch+1, not release-please's prediction**, which moves
mid-flight
and would reset `N` under already-published previews.
- **The version is applied to the CI working tree only**, never
committed —
`package.json` and the release-please manifest stay release-please's to
own.
- **`--tag preview` is mandatory:** `npm publish` defaults to `latest`
even for a
semver prerelease.
- **Three jobs.** `publish-npm-preview` mirrors `publish-npm` and only
publishes
(no repo write access); `publish-tag-preview` only tags, so a failed tag
is
retryable with _Re-run failed jobs_ without re-publishing an immutable
version;
`trigger-registry-update` is shared with the stable path. Both
downstream jobs
gate on a `published` output the publish step sets, not on job status —
the tag
push happens after the publish, so a red job can still mean npm has the
version.
## Next step
The publish is `npm publish --dry-run` for now. The `workflow_run`
trigger and
the `release` environment's protected-branch policy mean this job cannot
run off
the default branch at all, so merging this is the only way to exercise
the
trigger, the gate and the version calculation — and with `published`
unset, the
rehearsal neither tags nor touches the registry.
Once one run looks right, a follow-up drops `--dry-run` and adds
`echo "published=true" >> "$GITHUB_OUTPUT"` to that step. Nothing else
changes.
## Summary
- translate Claude SDK permission suggestions and tool-specific
decisions into ACP v1 presentation and editable option metadata
- align Claude session permission modes and clear-context plan handling
- split permission effects, modes, normalization, option construction,
presentation, and response handling into focused modules
- document the permission extension and add unit plus live-audit
coverage for native edge cases
## Tests
- `npm run check`
- `npm run build`
- `env -u ANTHROPIC_BASE_URL npm run test:run` (931 passed, 27 skipped)
## Summary
- align `_meta.jetbrains.air.sessionFailure` with AIR's
transcript-record protocol: `id`, `revision`, broad `category`, explicit
`severity`, `title`, optional `details`, and `actions`
- forward Claude's actual user-facing assistant/result text as `title`;
use adapter-authored fallback text only when Claude provides no message
(transport failure, worker EOF, or an internal invariant)
- reduce categories to `connection`, `access`, `limit`, `request`,
`service`, and `unknown`
- limit suggested actions to `retry`, `login`, and `new_session`
- update one transcript record through `id + revision`, keep resolved
entries in history, and remove clear/tombstone updates
- suppress duplicate assistant failure prose only for clients that
negotiated typed failures; preserve legacy ACP behavior otherwise
- restore synthetic usage-limit failures from session history with their
exact stored Claude text
- document negotiation, delivery carriers, identity/revision behavior,
mappings, recovery, and compatibility in the extension spec
## Validation
- `npm run check`
- `npm run build`
- `npm run test:run` — 777 passed, 26 skipped
- `git diff --check`
## Tracking
[IJAI-993](https://youtrack.jetbrains.com/issue/IJAI-993)
## Summary
- publish optimistic provider-neutral goal snapshots for submitted and
steered goal commands
- advertise and implement `set`, using steering for an active turn and
an ordinary prompt while idle
- suppress stale old-goal runtime updates while replacement/clear is
pending and roll back failed commands
- keep cancellation turn-scoped and orphan-account pending steers so
stale results cannot settle the next prompt
- reject unsupported pause/resume controls without host-side fallbacks
## Validation
- `npm test -- --run` (702 passed, 20 skipped)
- `npm run check`
- `npm run build`
- regression coverage for idle/active set and clear, stale runtime
updates, command failure rollback, cancellation, and stale steer
isolation
## Summary
- advertise a provider-neutral goal extension in
`InitializeResponse._meta.goal`
- map Claude SDK `active_goal` messages to
`session_info_update._meta.goal`
- expose the shared `_session/goal` control method with the truthfully
supported `clear` action
- document the extension as a candidate shape for a future first-class
ACP goal API
## Architecture
A goal is persistent **session state**, not the lifecycle of one
`session/prompt` request. `goal.status === "active"` says that the
objective may drive more work; it does not say that a prompt is
executing now.
This keeps two independent axes:
1. **Goal state** is published asynchronously as session metadata and
may survive multiple prompt turns.
2. **Prompt activity** follows the backend's actual running/quiescent
boundaries. While a turn runs, clients use advertised steering or prompt
queueing. Once the backend is quiescent, an ordinary `session/prompt` is
valid even if the goal remains active.
The metadata is deliberately provider-neutral:
```json
{
"_meta": {
"goal": {
"version": 1,
"controlMethod": "_session/goal",
"actions": ["clear"]
}
}
}
```
State is emitted as `session_info_update._meta.goal`; `null` clears it.
The common snapshot vocabulary supports `objective`, common statuses,
timestamps in Unix milliseconds, budgets/usage, iteration count, and a
continuation reason. Each implementation emits only fields it actually
knows. This shape is intentionally close to a future typed ACP
capability/update so promotion out of `_meta` would not require a
provider-specific migration.
Control actions are capability-negotiated. The shared vocabulary
reserves `set`, `pause`, `resume`, and `clear`, but an agent advertises
only the subset it can implement with correct semantics.
## Claude mapping
The Claude runtime already emits internal `active_goal` messages,
although that message is not currently included in the public
`SDKMessage` union. The adapter handles it before the exhaustive switch:
- `condition` → `objective`
- `iterations` → `iterations`
- `set_at` → `createdAt`
- `last_reason` → `lastReason`
- `value: null` → `goal: null`
Provider bookkeeping such as `tokens_at_start` is intentionally not
exposed. Goal updates do not settle, extend, or otherwise mutate ACP
prompt lifecycle. Claude's existing Stop hook remains authoritative for
whether the backend itself continues working.
Only `clear` is advertised for out-of-band control. It submits `/goal
clear` through the existing session prompt queue, preserving the
adapter's established ordering and turn ownership. Setting a goal
remains available through `/goal <condition>` but is not advertised as a
control action because it starts the long-running Stop-hook workflow.
The companion Codex implementation is agentclientprotocol/codex-acp#371.
## Verification
- `npm run test:run` — 692 passed, 20 skipped
- `npm run build`
- `npm run check`
Allow providing Claude model settings via env var on startup. Providing
model settings is required when using Claude with alternative llm
provider, e.g. Bedrock as models have different IDs there
nikita-ashihmin