feat: HITL continuation recording/replay support (#233)

## Summary

Closes #232. Adds `toolCallId`-based fixture matching so aimock can
record and replay HITL (Human-in-the-Loop) continuation requests — the
second leg of a `useHumanInTheLoop` flow where the last message is
`role: "tool"` instead of `role: "user"`.

- Add `toolCallId?: string` to `AGUIFixtureMatch` and
`AGUIConfigFixture`
- Add `getLastMessageIfToolResult` helper (checks absolute last message
only)
- Add `toolCallId` check in `matchesFixture` (AND logic with existing
fields)
- Recorder uses tool-result-first priority: `toolCallId` → user message
→ sentinel fallback
- Add `onToolResult(toolCallId, events, delayMs?)` fluent API method
- Config loader passes through `toolCallId` in events path, warns on
text shorthand misuse
- Export `getLastMessageIfToolResult` from both entry points

Also hardens the recorder against pre-existing robustness issues
surfaced during CR:
- Guard against recording fixtures from error (non-2xx) upstream
responses
- Add `settled` flag to prevent double-settle on error+end race
- Skip disk write for predicate fixtures (`__NO_USER_MESSAGE__` sentinel
was semantically broken on reload)
- Include parse error reason in SSE warning log

## Test plan

- 19 new tests (51 total), all passing
- Matcher: toolCallId match, wrong ID, non-tool message, absent
toolCallId, AND logic with message, AND logic with stateKey
- Fluent API: onToolResult registers and matches, onToolResult with
delayMs
- Recorder: continuation writes toolCallId fixture, continuation replay,
normal request still writes message, sentinel fallback, priority test
(tool result wins over user message in history)
- Integration: full HITL round-trip — record both legs, replay both
without upstream

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Author: jpr5

.claude-plugin/marketplace.json

@@ -9,7 +9,7 @@
       "source": {
         "source": "npm",
         "package": "@copilotkit/aimock",
-        "version": "^1.26.1"
+        "version": "^1.27.0"
       },
       "description": "Fixture authoring skill for @copilotkit/aimock — LLM, multimedia (image/TTS/transcription/video), MCP, A2A, AG-UI, vector, embeddings, structured output, sequential responses, streaming physics, record/replay, agent loop patterns, and debugging"
     }

.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "aimock",
-  "version": "1.26.1",
+  "version": "1.27.0",
   "description": "Fixture authoring guidance for @copilotkit/aimock — LLM, multimedia, MCP, A2A, AG-UI, vector, and service mocking",
   "author": {
     "name": "CopilotKit"

CHANGELOG.md

@@ -2,17 +2,16 @@
 
 ## [Unreleased]
 
-### Fixed
+## [1.27.0] - 2026-05-20
 
-- **AG-UI recorder** — `extractLastUserMessage` now walks structured `content`
-  arrays (e.g. `[{ type: "text", text: "..." }, { type: "document", source: ... }]`)
-  and joins their text parts. Previously, structured content fell back to the
-  `__NO_USER_MESSAGE__` sentinel, producing fixtures that couldn't replay.
+### Added
 
-### Changed
+- **HITL continuation recording/replay support** — `toolCallId` matching for continuation fixtures. New `toolCallId` field on `AGUIFixtureMatch` and `AGUIConfigFixture`. `getLastMessageIfToolResult` helper and `onToolResult` fluent API. Recorder uses tool-result-first priority for continuation fixtures. ([#233](https://github.com/CopilotKit/aimock/pull/233), closes [#232](https://github.com/CopilotKit/aimock/issues/232))
+
+### Fixed
 
-- **`AGUIMessage.content` type widened** to `string | AGUIMessageContentPart[]`.
-  New exported type `AGUIMessageContentPart` describes the per-part shape.
+- **Walk structured content arrays in `extractLastUserMessage`** — handle multimodal user content (`AGUIMessageContentPart[]`) by joining text parts and skipping non-text. Export `NO_USER_MESSAGE_SENTINEL` constant and `AGUIMessageContentPart` type. ([#231](https://github.com/CopilotKit/aimock/pull/231))
+- **Harden recorder against error responses, double-settle, and broken sentinel persistence** — guard against recording fixtures from non-2xx upstream responses, add `settled` flag to prevent error+end race, skip disk write for predicate fixtures (sentinel was semantically broken on reload), include parse error reason in SSE warning log
 
 ## [1.26.1] - 2026-05-19
 

charts/aimock/Chart.yaml

@@ -3,4 +3,4 @@ name: aimock
 description: Mock infrastructure for AI application testing (OpenAI, Anthropic, Gemini, MCP, A2A, vector)
 type: application
 version: 0.1.0
-appVersion: "1.26.1"
+appVersion: "1.27.0"

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@copilotkit/aimock",
-  "version": "1.26.1",
+  "version": "1.27.0",
   "description": "Mock infrastructure for AI application testing — LLM APIs, image generation, text-to-speech, transcription, audio generation, video generation, MCP tools, A2A agents, AG-UI event streams, vector databases, search, rerank, and moderation. One package, one port, zero dependencies.",
   "license": "MIT",
   "keywords": [