mirror of
https://github.com/agentclientprotocol/claude-agent-acp.git
synced 2026-11-07 18:01:07 +00:00
Compare commits
4 Commits
| Author | SHA1 | Date | |
|---|---|---|---|
|
d5e1b7fa72 | ||
|
|
960f62d765 | ||
|
|
508453c288 | ||
|
|
8dc8c86426 |
@@ -1,3 +1,3 @@
|
||||
{
|
||||
".": "0.48.0"
|
||||
".": "0.49.0"
|
||||
}
|
||||
|
||||
13
CHANGELOG.md
13
CHANGELOG.md
@@ -1,5 +1,18 @@
|
||||
# Changelog
|
||||
|
||||
## [0.49.0](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.48.0...v0.49.0) (2026-06-22)
|
||||
|
||||
|
||||
### Features
|
||||
|
||||
* Update to claude-agent-sdk 0.3.185 ([#798](https://github.com/agentclientprotocol/claude-agent-acp/issues/798)) ([8dc8c86](https://github.com/agentclientprotocol/claude-agent-acp/commit/8dc8c864263fa50d03bec7ebff7aee776604cb3d))
|
||||
|
||||
|
||||
### Bug Fixes
|
||||
|
||||
* Deduplicate streamed assistant blocks by content ([#800](https://github.com/agentclientprotocol/claude-agent-acp/issues/800)) ([960f62d](https://github.com/agentclientprotocol/claude-agent-acp/commit/960f62d76582ae5c9c5575ab66974809049ce1d0))
|
||||
* Infer 1M context from model descriptions ([#799](https://github.com/agentclientprotocol/claude-agent-acp/issues/799)) ([508453c](https://github.com/agentclientprotocol/claude-agent-acp/commit/508453c288b4a12701abd507199e7fa0ab172171))
|
||||
|
||||
## [0.48.0](https://github.com/agentclientprotocol/claude-agent-acp/compare/v0.47.0...v0.48.0) (2026-06-19)
|
||||
|
||||
|
||||
|
||||
@@ -40,7 +40,7 @@ export default [
|
||||
"default-case": "error",
|
||||
"prefer-const": "error",
|
||||
"no-var": "error",
|
||||
eqeqeq: ["error", "always"],
|
||||
eqeqeq: ["error", "smart"],
|
||||
curly: ["error", "all"],
|
||||
},
|
||||
},
|
||||
|
||||
88
package-lock.json
generated
88
package-lock.json
generated
@@ -1,16 +1,16 @@
|
||||
{
|
||||
"name": "@agentclientprotocol/claude-agent-acp",
|
||||
"version": "0.48.0",
|
||||
"version": "0.49.0",
|
||||
"lockfileVersion": 3,
|
||||
"requires": true,
|
||||
"packages": {
|
||||
"": {
|
||||
"name": "@agentclientprotocol/claude-agent-acp",
|
||||
"version": "0.48.0",
|
||||
"version": "0.49.0",
|
||||
"license": "Apache-2.0",
|
||||
"dependencies": {
|
||||
"@agentclientprotocol/sdk": "0.28.1",
|
||||
"@anthropic-ai/claude-agent-sdk": "0.3.183",
|
||||
"@anthropic-ai/claude-agent-sdk": "0.3.185",
|
||||
"zod": "^3.25.0 || ^4.0.0"
|
||||
},
|
||||
"bin": {
|
||||
@@ -41,22 +41,22 @@
|
||||
}
|
||||
},
|
||||
"node_modules/@anthropic-ai/claude-agent-sdk": {
|
||||
"version": "0.3.183",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk/-/claude-agent-sdk-0.3.183.tgz",
|
||||
"integrity": "sha512-rAKws76RKdFu+DK5q9wYN0AGLRdPrw0vGd+TMzSFynFdxwUAehYc9vUKJoK27j2xRoHFdNt+JDsih5BzsUzi+Q==",
|
||||
"version": "0.3.185",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk/-/claude-agent-sdk-0.3.185.tgz",
|
||||
"integrity": "sha512-UDDd0cInvOufmR88btR/Ursnh71sb5scoutNlRS2L0LBdkt7JEqYJ0jt7DnOYHEVhMx8AAcg0eYgvXztcWUnFA==",
|
||||
"license": "SEE LICENSE IN README.md",
|
||||
"engines": {
|
||||
"node": ">=18.0.0"
|
||||
},
|
||||
"optionalDependencies": {
|
||||
"@anthropic-ai/claude-agent-sdk-darwin-arm64": "0.3.183",
|
||||
"@anthropic-ai/claude-agent-sdk-darwin-x64": "0.3.183",
|
||||
"@anthropic-ai/claude-agent-sdk-linux-arm64": "0.3.183",
|
||||
"@anthropic-ai/claude-agent-sdk-linux-arm64-musl": "0.3.183",
|
||||
"@anthropic-ai/claude-agent-sdk-linux-x64": "0.3.183",
|
||||
"@anthropic-ai/claude-agent-sdk-linux-x64-musl": "0.3.183",
|
||||
"@anthropic-ai/claude-agent-sdk-win32-arm64": "0.3.183",
|
||||
"@anthropic-ai/claude-agent-sdk-win32-x64": "0.3.183"
|
||||
"@anthropic-ai/claude-agent-sdk-darwin-arm64": "0.3.185",
|
||||
"@anthropic-ai/claude-agent-sdk-darwin-x64": "0.3.185",
|
||||
"@anthropic-ai/claude-agent-sdk-linux-arm64": "0.3.185",
|
||||
"@anthropic-ai/claude-agent-sdk-linux-arm64-musl": "0.3.185",
|
||||
"@anthropic-ai/claude-agent-sdk-linux-x64": "0.3.185",
|
||||
"@anthropic-ai/claude-agent-sdk-linux-x64-musl": "0.3.185",
|
||||
"@anthropic-ai/claude-agent-sdk-win32-arm64": "0.3.185",
|
||||
"@anthropic-ai/claude-agent-sdk-win32-x64": "0.3.185"
|
||||
},
|
||||
"peerDependencies": {
|
||||
"@anthropic-ai/sdk": ">=0.93.0",
|
||||
@@ -65,9 +65,9 @@
|
||||
}
|
||||
},
|
||||
"node_modules/@anthropic-ai/claude-agent-sdk-darwin-arm64": {
|
||||
"version": "0.3.183",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk-darwin-arm64/-/claude-agent-sdk-darwin-arm64-0.3.183.tgz",
|
||||
"integrity": "sha512-o/+sxwKgXuw6RG5cERWjvcvL1CDSPe/TaXMhax+dq+V4lDOI5iTqg3y5Wfb6dL3xlWoTA2OhWowDQllKbE04LQ==",
|
||||
"version": "0.3.185",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk-darwin-arm64/-/claude-agent-sdk-darwin-arm64-0.3.185.tgz",
|
||||
"integrity": "sha512-P2Svxx1IFrS9aw4ylJBtcoJrc/OAOVnDs+o05x4TV7HoHUQUcKZ9XhWOGnadyVT0KLyoktgjyjnaK7Zdtc0Ldg==",
|
||||
"cpu": [
|
||||
"arm64"
|
||||
],
|
||||
@@ -78,9 +78,9 @@
|
||||
]
|
||||
},
|
||||
"node_modules/@anthropic-ai/claude-agent-sdk-darwin-x64": {
|
||||
"version": "0.3.183",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk-darwin-x64/-/claude-agent-sdk-darwin-x64-0.3.183.tgz",
|
||||
"integrity": "sha512-V7Cf8JeD5EPf4MPomFUlEblCIQI0wg+aWdOSqvfMsDmCBEHljd52CQ3a7W263oVt6I7QUfRTpX2KNvdma56rDA==",
|
||||
"version": "0.3.185",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk-darwin-x64/-/claude-agent-sdk-darwin-x64-0.3.185.tgz",
|
||||
"integrity": "sha512-H69m2hIJawzSkCDNO/CPLQCYvDGdlbZdzGTgGF8/IQuB4O81sv8WSAh6TBXwtFXOgEi4HtrUYlxiFDmTHt7hMA==",
|
||||
"cpu": [
|
||||
"x64"
|
||||
],
|
||||
@@ -91,9 +91,9 @@
|
||||
]
|
||||
},
|
||||
"node_modules/@anthropic-ai/claude-agent-sdk-linux-arm64": {
|
||||
"version": "0.3.183",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk-linux-arm64/-/claude-agent-sdk-linux-arm64-0.3.183.tgz",
|
||||
"integrity": "sha512-qqnSf85D3uBH7kRHLGOMhr7aQrREUZ4hUlCztrkJe/twUT4bmJsO86zamnAf7vOYeCul/sHYwBMmXtOfI+cHZg==",
|
||||
"version": "0.3.185",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk-linux-arm64/-/claude-agent-sdk-linux-arm64-0.3.185.tgz",
|
||||
"integrity": "sha512-KtXMCoprGYTSPb/A0/kjrO+PpDJAFbHhDd8rjqA4izU51OYjTDkahsGCcSTFM2jA8krOhPqzx/SNVuAVWmexwA==",
|
||||
"cpu": [
|
||||
"arm64"
|
||||
],
|
||||
@@ -107,9 +107,9 @@
|
||||
]
|
||||
},
|
||||
"node_modules/@anthropic-ai/claude-agent-sdk-linux-arm64-musl": {
|
||||
"version": "0.3.183",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk-linux-arm64-musl/-/claude-agent-sdk-linux-arm64-musl-0.3.183.tgz",
|
||||
"integrity": "sha512-ua6XaiCcYYKfd0CY9cjoyYnTQWI2GocpbP7Gwvmr+9taW/lTpQGVQH3FrF2VFKXVSt7AaVE6Z7WN9oPxcxl3VA==",
|
||||
"version": "0.3.185",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk-linux-arm64-musl/-/claude-agent-sdk-linux-arm64-musl-0.3.185.tgz",
|
||||
"integrity": "sha512-1GRpKYrg5foomW+qLrxm/k2R/5PEU7RAuEdX65HV6lFMJR3HsGj8qfBq4KCJHTS3r4YdVrFLvPQ/mUqbY6klIA==",
|
||||
"cpu": [
|
||||
"arm64"
|
||||
],
|
||||
@@ -123,9 +123,9 @@
|
||||
]
|
||||
},
|
||||
"node_modules/@anthropic-ai/claude-agent-sdk-linux-x64": {
|
||||
"version": "0.3.183",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk-linux-x64/-/claude-agent-sdk-linux-x64-0.3.183.tgz",
|
||||
"integrity": "sha512-GvfNp9XsKWFvr0EH3NgX2pqzHPYhFeqhOmVc4A+2e+GdxBMYUEa0ylfPLJItahQU9FkTJAw2oQNnzTHPPm1GxA==",
|
||||
"version": "0.3.185",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk-linux-x64/-/claude-agent-sdk-linux-x64-0.3.185.tgz",
|
||||
"integrity": "sha512-8U5wu6dNcyzRkwoae0Gu5ZHs6lRSu8CXqd9yrilX3fHR9kKICrm76bLu5q2auuSc/8dx6VL5ZXqSzFKIMp//zw==",
|
||||
"cpu": [
|
||||
"x64"
|
||||
],
|
||||
@@ -139,9 +139,9 @@
|
||||
]
|
||||
},
|
||||
"node_modules/@anthropic-ai/claude-agent-sdk-linux-x64-musl": {
|
||||
"version": "0.3.183",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk-linux-x64-musl/-/claude-agent-sdk-linux-x64-musl-0.3.183.tgz",
|
||||
"integrity": "sha512-wp+p+4xeZOW+h4InMKR7nIdGlgYftXyR+ErOVz7i2tMRcKuYMoc5jRIYjdJkuETMSr5nsRj0rDeO5gG7M6MGZQ==",
|
||||
"version": "0.3.185",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk-linux-x64-musl/-/claude-agent-sdk-linux-x64-musl-0.3.185.tgz",
|
||||
"integrity": "sha512-YAky0Oez+YYrlcas/xDiaCm9AUX0z1YwM+pnk+CFxEr/ICmc7MbiFsnEIXFLPHaox9bE5iOzw4LS4b/c1tIJRg==",
|
||||
"cpu": [
|
||||
"x64"
|
||||
],
|
||||
@@ -155,9 +155,9 @@
|
||||
]
|
||||
},
|
||||
"node_modules/@anthropic-ai/claude-agent-sdk-win32-arm64": {
|
||||
"version": "0.3.183",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk-win32-arm64/-/claude-agent-sdk-win32-arm64-0.3.183.tgz",
|
||||
"integrity": "sha512-cGyMC9YDsrCQ4av74dMVa2liOb5oLFX4+1SxwJGRCW7mBVBazcM6rg7ifWbd9XfkDhNPKOIfyiloipyMfoaE4g==",
|
||||
"version": "0.3.185",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk-win32-arm64/-/claude-agent-sdk-win32-arm64-0.3.185.tgz",
|
||||
"integrity": "sha512-pJEOaCLQQQWSyBeseR/GAn4eEp4WtMDP/o5yPMuuAsNWeoJXM7cY4j/N3KBE6GZofgnYdWEXILF2uw3SjbtINA==",
|
||||
"cpu": [
|
||||
"arm64"
|
||||
],
|
||||
@@ -168,9 +168,9 @@
|
||||
]
|
||||
},
|
||||
"node_modules/@anthropic-ai/claude-agent-sdk-win32-x64": {
|
||||
"version": "0.3.183",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk-win32-x64/-/claude-agent-sdk-win32-x64-0.3.183.tgz",
|
||||
"integrity": "sha512-h/XzbrSmXGroTk/FYKR6J4/8G9vDb1HUUUeNXeBGqGW1kppIiWPJKLRzjtSe0brVjADOKOT6tE5IHK0mV/1gBw==",
|
||||
"version": "0.3.185",
|
||||
"resolved": "https://registry.npmjs.org/@anthropic-ai/claude-agent-sdk-win32-x64/-/claude-agent-sdk-win32-x64-0.3.185.tgz",
|
||||
"integrity": "sha512-/Kf9XneSuIkSHNwrxOVY0XT7RovAEcx5nDASKzku4994cZz54otU5mT2jT6dwevDLEStfgts5x/A4z0AePwVTQ==",
|
||||
"cpu": [
|
||||
"x64"
|
||||
],
|
||||
@@ -2954,9 +2954,9 @@
|
||||
"license": "MIT"
|
||||
},
|
||||
"node_modules/nanoid": {
|
||||
"version": "3.3.13",
|
||||
"resolved": "https://registry.npmjs.org/nanoid/-/nanoid-3.3.13.tgz",
|
||||
"integrity": "sha512-sPdqC6ByMVVGvF1ynvvMo0/o+oD1VX7DaHhijt1bFgjvBkHBib4t49GoNDhf2NDta4oeUNlaGbSt5K7qjZ955Q==",
|
||||
"version": "3.3.15",
|
||||
"resolved": "https://registry.npmjs.org/nanoid/-/nanoid-3.3.15.tgz",
|
||||
"integrity": "sha512-y7Wygv/7mEOvxTuEQDB8StXdMRBWf1kR/tlhAzBRUFkB2jfcLOAxO/SHmOO2zgz1pVgK29/kyupn059/bCHdjA==",
|
||||
"dev": true,
|
||||
"funding": [
|
||||
{
|
||||
@@ -3366,9 +3366,9 @@
|
||||
"peer": true
|
||||
},
|
||||
"node_modules/semver": {
|
||||
"version": "7.8.4",
|
||||
"resolved": "https://registry.npmjs.org/semver/-/semver-7.8.4.tgz",
|
||||
"integrity": "sha512-rUCObTnP32Q08R2uuIrt7r9PlEonuTmtuXYcW6s5kjdlj3xbnwe+21yXptAUYcMAABLkYYTtnmzb3w3EDZfueA==",
|
||||
"version": "7.8.5",
|
||||
"resolved": "https://registry.npmjs.org/semver/-/semver-7.8.5.tgz",
|
||||
"integrity": "sha512-Y7/KDsb8LjooZpwaqGyulO6DQlksgCncchHGk+sZIY4SBvUocMBEFH5Ur1fI4dV+Jvl0w6cjvucaIi40puRioA==",
|
||||
"dev": true,
|
||||
"license": "ISC",
|
||||
"bin": {
|
||||
|
||||
@@ -3,7 +3,7 @@
|
||||
"publishConfig": {
|
||||
"access": "public"
|
||||
},
|
||||
"version": "0.48.0",
|
||||
"version": "0.49.0",
|
||||
"description": "An ACP-compatible coding agent powered by the Claude Agent SDK (TypeScript)",
|
||||
"main": "dist/lib.js",
|
||||
"types": "dist/lib.d.ts",
|
||||
@@ -61,7 +61,7 @@
|
||||
"license": "Apache-2.0",
|
||||
"dependencies": {
|
||||
"@agentclientprotocol/sdk": "0.28.1",
|
||||
"@anthropic-ai/claude-agent-sdk": "0.3.183",
|
||||
"@anthropic-ai/claude-agent-sdk": "0.3.185",
|
||||
"zod": "^3.25.0 || ^4.0.0"
|
||||
},
|
||||
"devDependencies": {
|
||||
|
||||
206
src/acp-agent.ts
206
src/acp-agent.ts
@@ -1024,12 +1024,17 @@ export class ClaudeAcpAgent {
|
||||
// (whose delta events don't carry it) can all be tagged with the same,
|
||||
// replay-stable id.
|
||||
let currentStreamMessageId: string | undefined;
|
||||
// Per-message-id record of which assistant content actually streamed live
|
||||
// via `stream_event` deltas, split by block type, so the consolidated
|
||||
// `assistant` message can drop the duplicate blocks that already reached the
|
||||
// client as chunks while still forwarding any that a non-streaming gateway
|
||||
const streamedTextIds = new Set<string>();
|
||||
const streamedThinkingIds = new Set<string>();
|
||||
// The text/thinking blocks that have actually streamed live as
|
||||
// `stream_event` deltas for the message the next consolidated `assistant`
|
||||
// will repeat, in stream order, each accumulated to its full streamed text.
|
||||
// The consolidated handler diffs each assembled block against these and
|
||||
// forwards only the un-streamed remainder — nothing if it streamed in full
|
||||
// (the common case), the whole block if it never streamed (a non-streaming
|
||||
// gateway), or just the tail if the stream was cut short mid-block. Matching
|
||||
// on content rather than the Anthropic message id makes dedupe robust to
|
||||
// gateways that don't carry a stable/matching id across the stream and the
|
||||
// consolidated message. Reset after each consolidated message consumes it.
|
||||
const streamedBlocks: { index: number; type: "text" | "thinking"; text: string }[] = [];
|
||||
// Stop reason accumulated for the active turn (result subtype, refusal,
|
||||
// max_tokens, …). Reset per turn; read when the turn settles at idle.
|
||||
let stopReason: StopReason = "end_turn";
|
||||
@@ -1041,12 +1046,14 @@ export class ClaudeAcpAgent {
|
||||
lastAssistantError = undefined;
|
||||
lastRefusalExplanation = null;
|
||||
compactionInProgress = false;
|
||||
// Do NOT reset currentStreamMessageId here. `message_start` sets it per
|
||||
// message; clearing it on turn activation drops the id for any block that
|
||||
// streams after a mid-message activation (the replayed user echo with
|
||||
// --replay-user-messages), so those ids never enter streamedTextIds and the
|
||||
// consolidated assistant message re-emits already-streamed text as a
|
||||
// duplicate. #785 stopped resetting streamedTextIds here but left this line.
|
||||
// Do NOT reset currentStreamMessageId or streamedBlocks here. Turn
|
||||
// activation can fire mid-message (the replayed user echo with
|
||||
// --replay-user-messages lands between a message's blocks); clearing the
|
||||
// streamed-content record on activation would drop the blocks that
|
||||
// streamed before the echo, so the consolidated assistant message would
|
||||
// re-emit them as duplicates. streamedBlocks is bounded instead by being
|
||||
// cleared when each consolidated message consumes it. #785 stopped
|
||||
// resetting the streamed-content tracking here but left this line.
|
||||
stopReason = "end_turn";
|
||||
session.accumulatedUsage = {
|
||||
inputTokens: 0,
|
||||
@@ -1682,21 +1689,52 @@ export class ClaudeAcpAgent {
|
||||
// it) can all be tagged with the same, replay-stable id.
|
||||
if (message.event.type === "message_start") {
|
||||
currentStreamMessageId = message.event.message.id || undefined;
|
||||
// A new top-level message starts: clear any streamed-content
|
||||
// residue from a prior message that never reached its
|
||||
// consolidated reset — a cancelled turn breaks out before the
|
||||
// reset, and the synthetic-auth/system/local-command paths
|
||||
// `break` early too. Block indices restart at 0 each message, so
|
||||
// leftover entries would otherwise collide with this message's
|
||||
// blocks and re-emit (or truncate) already-streamed text. Gated on
|
||||
// `parent_tool_use_id === null` so a subagent stream can't clear
|
||||
// the top-level record. Fires once, before any of this message's
|
||||
// blocks, so it doesn't disturb the mid-message turn-activation
|
||||
// path the way resetting on turn activation would.
|
||||
if (message.parent_tool_use_id === null) {
|
||||
streamedBlocks.length = 0;
|
||||
}
|
||||
}
|
||||
// Record that this top-level message id actually streamed text/
|
||||
// thinking, so the `assistant` case below knows its assembled blocks
|
||||
// are duplicates (filter them) rather than the only copy (forward
|
||||
// them). Gated on `parent_tool_use_id === null` so a subagent stream
|
||||
// can't attribute its content to the top-level message id.
|
||||
// Accumulate the text/thinking actually streamed live, so the
|
||||
// `assistant` case below can diff its assembled blocks against what
|
||||
// already reached the client as chunks and forward only the
|
||||
// remainder. Gated on `parent_tool_use_id === null` so a subagent
|
||||
// stream can't attribute its content to the top-level message.
|
||||
// Contiguous deltas of the same block (same index and type) extend
|
||||
// the current entry; anything else opens a new one.
|
||||
if (
|
||||
currentStreamMessageId &&
|
||||
message.parent_tool_use_id === null &&
|
||||
message.event.type === "content_block_delta"
|
||||
) {
|
||||
if (message.event.delta.type === "text_delta") {
|
||||
streamedTextIds.add(currentStreamMessageId);
|
||||
} else if (message.event.delta.type === "thinking_delta") {
|
||||
streamedThinkingIds.add(currentStreamMessageId);
|
||||
const delta = message.event.delta;
|
||||
const chunk =
|
||||
delta.type === "text_delta"
|
||||
? { type: "text" as const, text: delta.text }
|
||||
: delta.type === "thinking_delta"
|
||||
? { type: "thinking" as const, text: delta.thinking }
|
||||
: undefined;
|
||||
// Skip empty deltas (some gateways emit empty thinking chunks —
|
||||
// #793): appending "" is a no-op, but pushing a "" entry would
|
||||
// create a block the consolidated handler's `text.length > 0`
|
||||
// guard can never consume, stalling the diff cursor and
|
||||
// re-emitting the next block as a duplicate.
|
||||
if (chunk && chunk.text.length > 0) {
|
||||
const index = message.event.index;
|
||||
const last = streamedBlocks[streamedBlocks.length - 1];
|
||||
if (last && last.index === index && last.type === chunk.type) {
|
||||
last.text += chunk.text;
|
||||
} else {
|
||||
streamedBlocks.push({ index, type: chunk.type, text: chunk.text });
|
||||
}
|
||||
}
|
||||
}
|
||||
if (
|
||||
@@ -1907,41 +1945,66 @@ export class ClaudeAcpAgent {
|
||||
|
||||
let content: typeof message.message.content;
|
||||
if (message.type === "assistant" && message.parent_tool_use_id === null) {
|
||||
// Top-level assistant message: drop text/thinking blocks already
|
||||
// streamed live as chunks, and forward (as a fallback) any that were
|
||||
// not, so non-streaming gateways still deliver the final answer.
|
||||
const id = messageIdForGrouping(message);
|
||||
content = message.message.content.filter((item) => {
|
||||
// Non-text blocks (tool_use, etc.) always pass through; their own
|
||||
// dedupe (`toolUseCache`) collapses the streamed/assembled pair.
|
||||
// Top-level assistant message: each text/thinking block may have
|
||||
// already been streamed live as deltas. Diff each against what
|
||||
// streamed (`streamedBlocks`, in document order) and forward only
|
||||
// the un-streamed remainder — nothing if it streamed in full (the
|
||||
// common case), the whole block if it never streamed (a
|
||||
// non-streaming gateway), or just the tail if the stream was cut
|
||||
// short mid-block. `streamPos` walks the streamed blocks in step
|
||||
// with the assembled text/thinking blocks; tool_use and other
|
||||
// blocks pass through untouched (their own `toolUseCache` collapses
|
||||
// the streamed/assembled pair) without advancing it.
|
||||
const blocks = message.message.content;
|
||||
const kept: typeof blocks = [];
|
||||
let streamPos = 0;
|
||||
for (const item of blocks) {
|
||||
if (item.type !== "text" && item.type !== "thinking") {
|
||||
return true;
|
||||
kept.push(item);
|
||||
continue;
|
||||
}
|
||||
// Already delivered live as a chunk of this exact type — drop the
|
||||
// duplicate. Checked per type so streaming one (e.g. text) doesn't
|
||||
// suppress an un-streamed block of the other (e.g. thinking).
|
||||
const streamedLive =
|
||||
id !== undefined &&
|
||||
(item.type === "text" ? streamedTextIds : streamedThinkingIds).has(id);
|
||||
if (streamedLive) {
|
||||
return false;
|
||||
const full = item.type === "text" ? item.text : item.thinking;
|
||||
// Empty assembled blocks carry nothing (some gateways emit an
|
||||
// empty `thinking` block before the real text) — drop them.
|
||||
if (full.length === 0) {
|
||||
continue;
|
||||
}
|
||||
// Empty assembled blocks carry nothing (some gateways emit an empty
|
||||
// `thinking` block before the real text) — don't forward stray
|
||||
// empty chunks.
|
||||
const text = item.type === "text" ? item.text : item.thinking;
|
||||
if (text.length === 0) {
|
||||
return false;
|
||||
// A streamed block of the same type whose accumulated text is a
|
||||
// prefix of this one was already (at least partly) delivered as
|
||||
// chunks; consume it and forward only what's left. A non-empty
|
||||
// streamed text is required so an empty/aborted streamed block
|
||||
// doesn't swallow the assembled copy.
|
||||
const streamed = streamedBlocks[streamPos];
|
||||
if (
|
||||
streamed &&
|
||||
streamed.type === item.type &&
|
||||
streamed.text.length > 0 &&
|
||||
full.startsWith(streamed.text)
|
||||
) {
|
||||
streamPos++;
|
||||
const remainder = full.slice(streamed.text.length);
|
||||
if (remainder.length === 0) {
|
||||
continue;
|
||||
}
|
||||
// Overwrite in place with just the un-streamed tail (the
|
||||
// assembled message isn't read again after this) so the block
|
||||
// keeps its exact SDK type.
|
||||
if (item.type === "text") {
|
||||
item.text = remainder;
|
||||
} else {
|
||||
item.thinking = remainder;
|
||||
}
|
||||
kept.push(item);
|
||||
continue;
|
||||
}
|
||||
return true;
|
||||
});
|
||||
// The consolidated message is the last place this id is needed for
|
||||
// dedupe — drop it so the streamed-id sets stay bounded to in-flight
|
||||
// streams rather than growing for the session's whole life.
|
||||
if (id !== undefined) {
|
||||
streamedTextIds.delete(id);
|
||||
streamedThinkingIds.delete(id);
|
||||
// Not matched: never streamed (or the stream diverged from the
|
||||
// assembled text) — forward the block in full.
|
||||
kept.push(item);
|
||||
}
|
||||
content = kept;
|
||||
// Consumed: reset so the next message's blocks accumulate fresh and
|
||||
// the record stays bounded to the in-flight message.
|
||||
streamedBlocks.length = 0;
|
||||
} else if (message.type === "assistant") {
|
||||
// Subagent assistant message (`parent_tool_use_id !== null`). It is
|
||||
// never streamed live and its text/thinking is internal to the tool
|
||||
@@ -2652,19 +2715,27 @@ export class ClaudeAcpAgent {
|
||||
o.id === configId && typeof o.currentValue === "string" ? { ...o, currentValue: value } : o,
|
||||
);
|
||||
} else if (configId === "model") {
|
||||
// `ModelInfo.supportsAutoMode` is the canonical SDK signal for clamping
|
||||
// modes below; its `displayName`/`description` also let us infer the
|
||||
// context window for semantic aliases (e.g. `default`) whose ID alone
|
||||
// carries no "1m" token.
|
||||
const newModelInfo = session.modelInfos.find((m) => m.value === value);
|
||||
if (session.models.currentModelId !== value) {
|
||||
// The cached context window was learned for the previous model; reset
|
||||
// to the new model's heuristic so mid-stream updates between now and
|
||||
// the next `result` reflect the user's selection instead of the old
|
||||
// model's window.
|
||||
session.contextWindowSize = inferContextWindowFromModel(value) ?? DEFAULT_CONTEXT_WINDOW;
|
||||
session.contextWindowSize =
|
||||
inferContextWindowFromModel(
|
||||
value,
|
||||
newModelInfo?.displayName,
|
||||
newModelInfo?.description,
|
||||
) ?? DEFAULT_CONTEXT_WINDOW;
|
||||
}
|
||||
session.models = { ...session.models, currentModelId: value };
|
||||
|
||||
// Recompute availableModes for the new model and clamp the current
|
||||
// mode if the SDK no longer offers it (today: "auto" on Haiku).
|
||||
// `ModelInfo.supportsAutoMode` is the canonical SDK signal.
|
||||
const newModelInfo = session.modelInfos.find((m) => m.value === value);
|
||||
const newAvailableModes = buildAvailableModes(newModelInfo);
|
||||
// Capture BEFORE mutating session.modes so the log message reflects
|
||||
// the invalidated mode rather than "default".
|
||||
@@ -3219,7 +3290,11 @@ export class ClaudeAcpAgent {
|
||||
abortController,
|
||||
emitRawSDKMessages: sessionMeta?.claudeCode?.emitRawSDKMessages ?? false,
|
||||
contextWindowSize:
|
||||
inferContextWindowFromModel(models.currentModelId) ?? DEFAULT_CONTEXT_WINDOW,
|
||||
inferContextWindowFromModel(
|
||||
models.currentModelId,
|
||||
currentModelInfo?.displayName,
|
||||
currentModelInfo?.description,
|
||||
) ?? DEFAULT_CONTEXT_WINDOW,
|
||||
taskState,
|
||||
toolUseCache: {},
|
||||
messageIdToUuid: new Map(),
|
||||
@@ -4377,14 +4452,21 @@ function commonPrefixLength(a: string, b: string) {
|
||||
return i;
|
||||
}
|
||||
|
||||
/** Best-effort first guess of a model's context window from its ID, used only
|
||||
* as a fallback when the SDK's authoritative `getContextUsage` is unavailable
|
||||
* (and until a `result` message arrives with the `modelUsage` value).
|
||||
/** Best-effort first guess of a model's context window, used only as a
|
||||
* fallback when the SDK's authoritative `getContextUsage` is unavailable (and
|
||||
* until a `result` message arrives with the `modelUsage` value).
|
||||
*
|
||||
* Anthropic 1M-context variants encode "1m" as a distinct token in the SDK
|
||||
* model ID (e.g., "claude-opus-4-6-1m"), which `\b1m\b` catches without also
|
||||
* matching things like "10m" or embedded substrings. */
|
||||
function inferContextWindowFromModel(model: string): number | null {
|
||||
if (/\b1m\b/i.test(model)) return 1_000_000;
|
||||
* matching things like "10m" or embedded substrings. Semantic aliases like
|
||||
* `default` carry no such token in the ID, but the SDK's human-facing
|
||||
* `displayName`/`description` do (e.g. "Opus 4.7 (1M context)"), so callers
|
||||
* pass those too — the same `\b1m\b` token appears in "1M context". The SDK's
|
||||
* `ModelInfo` exposes no structured context-window field, so this text scan is
|
||||
* the only pre-`result` signal available. A miss falls back to the default
|
||||
* window and is corrected by `result.modelUsage` within one turn. */
|
||||
function inferContextWindowFromModel(...texts: Array<string | undefined>): number | null {
|
||||
if (texts.some((text) => text != null && /\b1m\b/i.test(text))) return 1_000_000;
|
||||
return null;
|
||||
}
|
||||
|
||||
|
||||
@@ -3072,6 +3072,26 @@ describe("usage_update computation", () => {
|
||||
expect(usageUpdates[1].update.size).toBe(1000000);
|
||||
});
|
||||
|
||||
it("infers the 1M window from a model's description when the ID lacks a 1m token (issue #596)", async () => {
|
||||
// Semantic aliases like `default` resolve to a 1M-context model but carry
|
||||
// no "1m" token in the modelId — the SDK signals 1M only via the
|
||||
// human-facing displayName/description (e.g. "Opus 4.7 with 1M context").
|
||||
// Inference must read those so the session reports the correct window from
|
||||
// the first mid-stream update instead of the 200k placeholder.
|
||||
const { agent } = createMockAgentWithCapture();
|
||||
injectSession(agent, [{ type: "system", subtype: "session_state_changed", state: "idle" }]);
|
||||
const session = agent.sessions["test-session"];
|
||||
session.models = { currentModelId: "claude-sonnet-4-6", availableModels: [] };
|
||||
session.modelInfos = [
|
||||
{ value: "default", displayName: "Default", description: "Opus 4.7 with 1M context" },
|
||||
] as any;
|
||||
expect(session.contextWindowSize).toBe(200000);
|
||||
|
||||
await (agent as any).applyConfigOptionValue("test-session", session, "model", "default");
|
||||
|
||||
expect(session.contextWindowSize).toBe(1000000);
|
||||
});
|
||||
|
||||
it("result with no matching modelUsage preserves the learned window", async () => {
|
||||
// A turn whose `result.modelUsage` doesn't contain the current top-level
|
||||
// model (e.g. no top-level assistant message, or only a subagent ran) must
|
||||
@@ -3872,6 +3892,134 @@ describe("assembled assistant text fallback", () => {
|
||||
expect(thoughtChunkTexts(updates)).toEqual(["private reasoning"]);
|
||||
});
|
||||
|
||||
it("forwards only the un-streamed remainder when the stream is cut short mid-block", async () => {
|
||||
const { agent, updates } = createMockAgentWithCapture();
|
||||
// The stream stops partway ("hello ") but the consolidated message carries
|
||||
// the whole block ("hello world"). The streamed prefix must not be re-sent,
|
||||
// and the un-streamed tail must still reach the client — dropping the whole
|
||||
// assembled block would truncate the answer to "hello ".
|
||||
injectSession(agent, [
|
||||
messageStart("msg-partial"),
|
||||
textDelta("hello "),
|
||||
assistantMessage("msg-partial", [{ type: "text", text: "hello world" }]),
|
||||
result(),
|
||||
idle,
|
||||
]);
|
||||
|
||||
await agent.prompt({ sessionId: "test-session", prompt: [{ type: "text", text: "hi" }] });
|
||||
|
||||
// The streamed prefix, then just the tail from the consolidated message.
|
||||
expect(messageChunkTexts(updates)).toEqual(["hello ", "world"]);
|
||||
});
|
||||
|
||||
it("dedupes streamed text even when the consolidated message carries a different id", async () => {
|
||||
const { agent, updates } = createMockAgentWithCapture();
|
||||
// Some gateways assign one id during the stream and a different one (or only
|
||||
// a uuid) on the assembled message. Dedupe must key on content, not the id,
|
||||
// or the consolidated block re-emits already-streamed text as a duplicate.
|
||||
injectSession(agent, [
|
||||
messageStart("msg-stream-id"),
|
||||
textDelta("hello "),
|
||||
textDelta("world"),
|
||||
assistantMessage("msg-DIFFERENT-id", [{ type: "text", text: "hello world" }]),
|
||||
result(),
|
||||
idle,
|
||||
]);
|
||||
|
||||
await agent.prompt({ sessionId: "test-session", prompt: [{ type: "text", text: "hi" }] });
|
||||
|
||||
// Only the streamed deltas — the assembled copy is deduped despite the id
|
||||
// mismatch.
|
||||
expect(messageChunkTexts(updates)).toEqual(["hello ", "world"]);
|
||||
});
|
||||
|
||||
it("dedupes a streamed text block even when an empty thinking delta precedes it", async () => {
|
||||
// An empty thinking delta (some gateways emit them — #793) must not create
|
||||
// a zero-length streamedBlocks entry: that entry can never satisfy the
|
||||
// consolidated handler's `text.length > 0` guard, so it would stall the
|
||||
// diff cursor and re-emit the real, already-streamed text as a duplicate.
|
||||
const { agent, updates } = createMockAgentWithCapture();
|
||||
injectSession(agent, [
|
||||
messageStart("msg-empty-thinking"),
|
||||
thinkingDelta(""),
|
||||
textDelta("real answer"),
|
||||
assistantMessage("msg-empty-thinking", [{ type: "text", text: "real answer" }]),
|
||||
result(),
|
||||
idle,
|
||||
]);
|
||||
|
||||
await agent.prompt({ sessionId: "test-session", prompt: [{ type: "text", text: "hi" }] });
|
||||
|
||||
// The streamed text appears once; the consolidated copy is deduped.
|
||||
expect(messageChunkTexts(updates)).toEqual(["real answer"]);
|
||||
});
|
||||
|
||||
it("does not re-emit the next turn's text after a turn is cancelled mid-stream", async () => {
|
||||
// Regression: streamedBlocks is reset inside the consolidated-assistant
|
||||
// branch, but a cancelled turn `break`s out before reaching it (the
|
||||
// `if (session.cancelled) break;` guard), and streamedBlocks is
|
||||
// session-scoped — so a cancelled turn's streamed text used to leak into
|
||||
// the next turn. Block indices restart at 0 per message, so the leftover
|
||||
// "Hello there" would fuse with turn 2's first block and make its
|
||||
// consolidated copy fail the prefix dedupe, re-emitting "Second answer" as
|
||||
// a duplicate. The fix resets streamedBlocks on each top-level
|
||||
// `message_start`, bounding the record to one in-flight message.
|
||||
const { agent, updates } = createMockAgentWithCapture();
|
||||
|
||||
let releaseCancel!: () => void;
|
||||
const cancelled = new Promise<void>((resolve) => {
|
||||
releaseCancel = resolve;
|
||||
});
|
||||
|
||||
injectGeneratorSession(agent, (input) => {
|
||||
async function* messageGenerator() {
|
||||
const iter = input[Symbol.asyncIterator]();
|
||||
const u1 = await iter.next();
|
||||
yield userEcho(u1.value); // activate turn 1
|
||||
yield messageStart("msg-1");
|
||||
yield textDelta("Hello ");
|
||||
yield textDelta("there"); // streamedBlocks = [{ index: 0, text: "Hello there" }]
|
||||
await cancelled; // hold until the test has cancelled turn 1
|
||||
// Turn 1's consolidated message arrives while cancelled → hits the
|
||||
// `if (session.cancelled) break;` guard, skipping the streamedBlocks
|
||||
// reset. The leftover entry must not survive into turn 2.
|
||||
yield assistantMessage("msg-1", [{ type: "text", text: "Hello there" }]);
|
||||
yield idle; // settles turn 1 as cancelled
|
||||
const u2 = await iter.next();
|
||||
yield userEcho(u2.value); // activate turn 2
|
||||
yield messageStart("msg-2"); // resets streamedBlocks (the fix)
|
||||
yield textDelta("Second answer");
|
||||
yield assistantMessage("msg-2", [{ type: "text", text: "Second answer" }]);
|
||||
yield result();
|
||||
yield idle;
|
||||
}
|
||||
return messageGenerator();
|
||||
});
|
||||
|
||||
const first = agent.prompt({
|
||||
sessionId: "test-session",
|
||||
prompt: [{ type: "text", text: "first" }],
|
||||
});
|
||||
// Wait until turn 1's deltas have streamed before cancelling.
|
||||
const deadline = Date.now() + 1000;
|
||||
while (!messageChunkTexts(updates).includes("there")) {
|
||||
if (Date.now() > deadline) throw new Error("turn 1 stream never arrived");
|
||||
await new Promise((r) => setTimeout(r, 1));
|
||||
}
|
||||
|
||||
await agent.cancel({ sessionId: "test-session" });
|
||||
releaseCancel();
|
||||
await expect(first).resolves.toEqual({ stopReason: "cancelled" });
|
||||
|
||||
await agent.prompt({ sessionId: "test-session", prompt: [{ type: "text", text: "second" }] });
|
||||
|
||||
// Turn 2's text appears exactly once (the live delta); the consolidated copy
|
||||
// is deduped despite the cancelled turn's leftover streamed text.
|
||||
expect(messageChunkTexts(updates).filter((t) => t === "Second answer")).toEqual([
|
||||
"Second answer",
|
||||
]);
|
||||
});
|
||||
|
||||
it("does not leak subagent assistant text into the top-level feed", async () => {
|
||||
const { agent, updates } = createMockAgentWithCapture();
|
||||
// Subagent assistant messages (parent_tool_use_id !== null) are never
|
||||
|
||||
Reference in New Issue
Block a user