feat: task_completed signal — post-task skill capture (v0.3.2)

Adds an 11th signal type emitted when a run of work (between two
UserPromptSubmit events) crosses three quality gates:
  - >=5 tool calls (TASK_TOOL_MIN)
  - >=3 distinct tool kinds (TASK_DIVERSITY_MIN, filters single-tool
    sweeps like "wrote 5 files")
  - 0 correction signals during the run (filters tasks where the user
    pushed back; correction-during-task disqualifies the recipe)

Payload carries tool_count, tool_kinds, active_skills, active_agents
so the agent can cluster by sorted tool-kind tuple and route through
the existing skill-overlap rule (skill_new vs skill_edit).

Importantly: cross_session_evidence is FALSE on first occurrence,
so resulting skill_new proposals always queue for review — they only
auto-apply when the same multi-tool recipe recurs in a second session
(then the existing rubric kicks in). Post-task creation captures novel
patterns while preserving the rule "auto-apply requires cross-session".

Hook adds state fields: task_tool_count, task_tool_kinds, task_corrections.
All reset on UserPromptSubmit boundary and on session change.

Agent gets one new signal-types-table row and one clustering bullet
referencing the existing skill-overlap rule.

3 new tests (30 passed, 0 failed):
  - 5 tools + 5 kinds + 0 corrections fires task_completed
  - 5 tools + 1 kind (Edit only) does NOT fire (diversity gate)
  - 5 tools + 3 kinds + correction-on-closing-prompt does NOT fire

adam/tests/run-tests.sh

@@ -364,6 +364,48 @@ for i in 1 2 3 4 5; do
 done
 assert_grep "$ROOT/journal.jsonl" '"active_skills":\["caveman"\]' "active_skills payload includes invoked skill"
 
+# --- Test 24: task_completed fires on diverse multi-tool task ---
+echo "Test 24: task_completed after 5 tools / 3 kinds / no corrections"
+reset_state
+for kind in Bash Read Edit Write Grep; do
+  echo "{\"hook_event_name\":\"PostToolUse\",\"tool_name\":\"$kind\",\"tool_input\":{},\"session_id\":\"sT\",\"cwd\":\"/tmp/x\"}" \
+    | HOOK_RUN >/dev/null 2>&1 || true
+done
+echo '{"hook_event_name":"UserPromptSubmit","prompt":"go on","session_id":"sT","cwd":"/tmp/x"}' \
+  | HOOK_RUN >/dev/null 2>&1 || true
+assert_grep "$ROOT/journal.jsonl" '"type":"task_completed"' "5 tools + 5 kinds + 0 corrections emits task_completed"
+
+# --- Test 25: task_completed suppressed when tool diversity < 3 ---
+echo "Test 25: task_completed suppressed on single-tool run"
+reset_state
+for i in 1 2 3 4 5; do
+  echo "{\"hook_event_name\":\"PostToolUse\",\"tool_name\":\"Edit\",\"tool_input\":{\"file_path\":\"/tmp/$i\"},\"session_id\":\"sT2\",\"cwd\":\"/tmp/x\"}" \
+    | HOOK_RUN >/dev/null 2>&1 || true
+done
+echo '{"hook_event_name":"UserPromptSubmit","prompt":"go on","session_id":"sT2","cwd":"/tmp/x"}' \
+  | HOOK_RUN >/dev/null 2>&1 || true
+if grep -qE '"type":"task_completed"' "$ROOT/journal.jsonl"; then
+  echo "  FAIL: task_completed fired on single-tool task"; FAIL=$((FAIL+1))
+else
+  echo "  PASS: task_completed suppressed (low tool diversity)"; PASS=$((PASS+1))
+fi
+
+# --- Test 26: task_completed suppressed when correction fires mid-task ---
+echo "Test 26: task_completed suppressed after correction"
+reset_state
+for kind in Bash Read Edit Write Grep; do
+  echo "{\"hook_event_name\":\"PostToolUse\",\"tool_name\":\"$kind\",\"tool_input\":{},\"session_id\":\"sT3\",\"cwd\":\"/tmp/x\"}" \
+    | HOOK_RUN >/dev/null 2>&1 || true
+done
+# Correction phrase resets task_corrections inside the same UserPromptSubmit cycle, so the prior run is disqualified.
+echo '{"hook_event_name":"UserPromptSubmit","prompt":"no, undo that","session_id":"sT3","cwd":"/tmp/x"}' \
+  | HOOK_RUN >/dev/null 2>&1 || true
+if grep -qE '"type":"task_completed"' "$ROOT/journal.jsonl"; then
+  echo "  FAIL: task_completed fired despite correction on the closing prompt"; FAIL=$((FAIL+1))
+else
+  echo "  PASS: task_completed suppressed by correction"; PASS=$((PASS+1))
+fi
+
 echo
 echo "Results: $PASS passed, $FAIL failed"
 [ "$FAIL" = "0" ]

agents/adam.md

@@ -38,6 +38,7 @@ The hook emits these `type` values into the journal:
 | `subagent_dispatch_pattern` | same subagent dispatched ≥3× cumulatively | subagent_type |
 | `correction_free_streak` | 5 clean UserPromptSubmits in a row (no correction phrase) | `active_skills[0]` |
 | `clean_recovery` | 3 clean PostToolUse events after a `tool_error_loop`/`dead_end`/`retry_loop` | (`recovered_from`, `active_skills[0]`) |
+| `task_completed` | UserPromptSubmit closes a run of ≥5 tool calls with ≥3 distinct tool kinds and 0 corrections | sorted `tool_kinds` tuple |
 
 ## Process
 
@@ -62,6 +63,7 @@ The hook emits these `type` values into the journal:
    - `subagent_dispatch_pattern`: cluster by `subagent_type`.
    - `correction_free_streak`: cluster by `active_skills[0]`. Treat ≥3 streaks across ≥2 sessions naming the same skill as cross-session evidence.
    - `clean_recovery`: cluster by (`recovered_from`, `active_skills[0]`). A win cluster qualifies for `skill_edit` only when the named skill exists in `skills_root`.
+   - `task_completed`: cluster by sorted `tool_kinds` tuple (the multi-tool recipe). Single entry qualifies for `skill_new` proposal (drafting protocol applies). Cross-session evidence requires ≥2 entries from distinct sessions with same tuple — without it, proposal queues, never auto-applies. Run the existing skill-overlap rule before drafting: if the recipe matches an existing skill's name/description tokens, route to `skill_edit` instead.
 5. **Multi-axis correlation**: for each session that produced ≥2 distinct struggle types (`tool_error_loop`, `dead_end`, `weak_agent`, `retry_loop`, `edit_churn`, `build_loop`), tag clusters from that session as `multi_axis: true`. This grants +1 confidence at scoring.
 6. For each cluster qualifying under the rubric — ≥3 occurrences across ≥2 sessions, OR (for struggle types) ≥1 entry within a single session, OR (for `correction`) ≥3 occurrences across ≥2 cwds:
    a. If cluster topic matches a rejected idea via the rejected-ideas fuzzy set (≥2 token overlap with rejection's `# Why`), skip with reason `"rejected-similar"`.

hooks/adam-observe.mjs

@@ -32,6 +32,8 @@ const CORRECTION_FREE_THRESHOLD = 5;
 const CLEAN_RECOVERY_WINDOW = 3;
 const STRUGGLE_TYPES = new Set(["tool_error_loop", "dead_end", "retry_loop"]);
 const ACTIVE_SKILLS_LOOKBACK = 10;
+const TASK_TOOL_MIN = 5;
+const TASK_DIVERSITY_MIN = 3;
 const STATE_MAX_BYTES = 1_000_000;
 
 function safeRead(path, fallback) {
@@ -133,6 +135,9 @@ function resetSessionLocal(state) {
   state.correctionFreeCounter = 0;
   state.recoveryWatch = null;
   state.tool_window = [];
+  state.task_tool_kinds = {};
+  state.task_tool_count = 0;
+  state.task_corrections = 0;
 }
 
 function ensureStateDefaults(state) {
@@ -149,6 +154,9 @@ function ensureStateDefaults(state) {
   if (typeof state.correctionFreeCounter !== "number") state.correctionFreeCounter = 0;
   if (state.recoveryWatch === undefined) state.recoveryWatch = null;
   if (!Array.isArray(state.activity_ring)) state.activity_ring = [];
+  if (!state.task_tool_kinds || typeof state.task_tool_kinds !== "object") state.task_tool_kinds = {};
+  if (typeof state.task_tool_count !== "number") state.task_tool_count = 0;
+  if (typeof state.task_corrections !== "number") state.task_corrections = 0;
 }
 
 function main() {
@@ -178,6 +186,7 @@ function main() {
         prev_file: last.file || null,
       });
       state.correctionFreeCounter = 0;
+      state.task_corrections += 1;
     } else {
       state.correctionFreeCounter += 1;
       if (state.correctionFreeCounter >= CORRECTION_FREE_THRESHOLD) {
@@ -190,6 +199,22 @@ function main() {
         state.correctionFreeCounter = 0;
       }
     }
+    // Evaluate prior task (work between previous UserPromptSubmit and this one).
+    const taskKinds = Object.keys(state.task_tool_kinds);
+    if (state.task_tool_count >= TASK_TOOL_MIN &&
+        taskKinds.length >= TASK_DIVERSITY_MIN &&
+        state.task_corrections === 0) {
+      appendJournal({
+        ts, session, cwd, type: "task_completed",
+        tool_count: state.task_tool_count,
+        tool_kinds: taskKinds,
+        active_skills: activeNames(state, "skill"),
+        active_agents: activeNames(state, "agent"),
+      });
+    }
+    state.task_tool_kinds = {};
+    state.task_tool_count = 0;
+    state.task_corrections = 0;
     resetFrictionCounters(state);
   } else if (event === "PreToolUse") {
     const tool = input.tool_name;
@@ -293,6 +318,9 @@ function main() {
       state.dead_end_emitted = true;
     }
 
+    state.task_tool_count += 1;
+    state.task_tool_kinds[tool] = (state.task_tool_kinds[tool] || 0) + 1;
+
     if (struggleEmittedThisTurn) {
       state.recoveryWatch = { recovered_from: struggleEmittedThisTurn, since_ts: ts, clean_count: 0, window_tools: [] };
     } else if (state.recoveryWatch) {