@event4u/agent-config 8.1.0 → 8.3.0

package/tests/fixtures/hooks/README.md

@@ -0,0 +1,92 @@
+# Hook fixture corpus
+
+**Purpose.** Minimal, valid, platform-native payloads — one per event in
+the agent-config event vocabulary
+(see [`hook-architecture-v1`](../../../docs/contracts/hook-architecture-v1.md)).
+Consumed by `./agent-config hooks:replay` (Phase 2.4b of
+`agents/roadmaps/road-to-proof-not-features.md`) and by the dispatcher
+replay tests (Phase 2.4c).
+
+**Layout.** One file per event:
+
+```
+tests/fixtures/hooks/
+  session_start.json     — augment-shaped startup envelope
+  session_end.json       — augment-shaped end envelope
+  user_prompt_submit.json — claude-shaped (augment has no native event)
+  pre_tool_use.json      — claude-shaped (augment optional)
+  post_tool_use.json     — augment-shaped tool-call envelope
+  stop.json              — augment-shaped end-of-turn envelope
+  pre_compact.json       — claude-shaped compaction envelope
+  agent_error.json       — synthetic (dispatcher contract § Round 2 Q3)
+```
+
+Each fixture is a **stdin payload**, not a dispatcher envelope. The
+dispatcher wraps the payload via `_build_envelope` before handing it
+to a concern (see `scripts/hooks/dispatch_hook.py`).
+
+## Schema invariants
+
+Every fixture MUST:
+
+1. Be valid JSON (`json.loads` round-trips).
+2. Be an object at the top level (the dispatcher only wraps dicts).
+3. Carry `session_id` (string, non-empty) so feedback writes land in a
+   deterministic slot under `agents/runtime/state/.dispatcher/<session_id>/`.
+4. Carry enough event-specific fields that real concerns
+   (`chat-history`, `roadmap-progress`, `context-hygiene`,
+   `verify-before-complete`, `minimal-safe-diff`) can run without
+   raising — primarily `tool_name` (for `*_tool_use`) and `prompt` /
+   `text` (for `user_prompt_submit`).
+5. Carry **no** real user content. Use placeholder strings
+   (`"hello"`, `"echo hi"`) — these files are committed; redaction is
+   not enforced post-commit.
+
+## Choosing the platform shape per event
+
+Picked one representative shape per event (rather than N × M fixtures)
+so the corpus stays small and the matching test grid stays readable.
+Multi-platform shape coverage lives in
+`tests/hooks/test_event_shape_contract.py` (frozen samples table).
+
+| Event | Shape | Why |
+|---|---|---|
+| `session_start` | augment | augment ships in this repo's primary CI |
+| `session_end` | augment | symmetry with `session_start` |
+| `user_prompt_submit` | claude | augment has no native event for it |
+| `pre_tool_use` | claude | optional on augment, native on claude |
+| `post_tool_use` | augment | most-bound event, all platforms support |
+| `stop` | augment | all platforms bind it (except copilot) |
+| `pre_compact` | claude | claude-only event |
+| `agent_error` | synthetic | dispatcher-defined; no native source |
+
+## Usage
+
+```bash
+# Replay one fixture through the dispatcher (read-only, replay mode):
+AGENT_CONFIG_REPLAY=1 \
+  python3 scripts/hooks/dispatch_hook.py \
+    --platform augment --event post_tool_use \
+    < tests/fixtures/hooks/post_tool_use.json
+
+# Or via the CLI subcommand (Phase 2.4b):
+./agent-config hooks:replay \
+  --platform augment --event post_tool_use \
+  --payload tests/fixtures/hooks/post_tool_use.json
+```
+
+`AGENT_CONFIG_REPLAY=1` signals concerns to skip writes under
+`agents/runtime/state/`. Concerns that don't honor the flag are listed by
+`./agent-config hooks:doctor` as not replay-safe.
+
+## When to update
+
+- Vendor schema drift surfaces in `test_event_shape_contract.py` first
+  (frozen samples table). When a sample changes, update the matching
+  fixture here to keep the corpus aligned.
+- New event added to `EVENT_VOCABULARY` in
+  `scripts/hooks/dispatch_hook.py` → add a fixture here. The replay
+  test (Phase 2.4c) asserts a 1:1 mapping between the vocabulary set
+  and this directory.
+- **Never** edit a fixture to make a failing test pass. The fixture is
+  the contract; the test follows it.

package/tests/fixtures/hooks/agent_error.json

@@ -0,0 +1,7 @@
+{
+  "session_id": "fixture-agent-error-001",
+  "error_type": "host_crash",
+  "error_message": "subprocess terminated unexpectedly",
+  "cwd": "/workspace",
+  "platform_hint": "synthetic"
+}

package/tests/fixtures/hooks/post_tool_use.json

@@ -0,0 +1,13 @@
+{
+  "session_id": "fixture-post-tool-001",
+  "tool_name": "str-replace-editor",
+  "tool_input": {
+    "path": "agents/roadmaps/example.md",
+    "command": "str_replace"
+  },
+  "tool_response": {
+    "success": true
+  },
+  "cwd": "/workspace",
+  "platform_hint": "augment"
+}

package/tests/fixtures/hooks/pre_compact.json

@@ -0,0 +1,7 @@
+{
+  "session_id": "fixture-pre-compact-001",
+  "trigger": "auto",
+  "transcript_path": "/tmp/fixture-transcript.json",
+  "cwd": "/workspace",
+  "platform_hint": "claude"
+}

package/tests/fixtures/hooks/pre_tool_use.json

@@ -0,0 +1,9 @@
+{
+  "session_id": "fixture-pre-tool-001",
+  "tool_name": "Read",
+  "tool_input": {
+    "file_path": "/workspace/README.md"
+  },
+  "cwd": "/workspace",
+  "platform_hint": "claude"
+}

package/tests/fixtures/hooks/session_end.json

@@ -0,0 +1,6 @@
+{
+  "session_id": "fixture-session-end-001",
+  "reason": "user_quit",
+  "cwd": "/workspace",
+  "platform_hint": "augment"
+}

package/tests/fixtures/hooks/session_start.json

@@ -0,0 +1,6 @@
+{
+  "session_id": "fixture-session-start-001",
+  "source": "startup",
+  "cwd": "/workspace",
+  "platform_hint": "augment"
+}

package/tests/fixtures/hooks/stop.json

@@ -0,0 +1,6 @@
+{
+  "session_id": "fixture-stop-001",
+  "stop_reason": "end_turn",
+  "cwd": "/workspace",
+  "platform_hint": "augment"
+}

package/tests/fixtures/hooks/user_prompt_submit.json

@@ -0,0 +1,7 @@
+{
+  "session_id": "fixture-user-prompt-001",
+  "prompt": "echo hello",
+  "transcript_path": "/tmp/fixture-transcript.json",
+  "cwd": "/workspace",
+  "platform_hint": "claude"
+}