Closes brain-retro 2026-05-20 #18 — episodes without schema_version=2
(legacy v1 era pre-2026-05-19T08:06) are now visible in STATUS.md
metrics. They're already filtered out of factor analysis by analyzer's
v1SkippedCount, but their existence was invisible to humans reading
STATUS — masking the bootstrap-epoch gap.
2 new vitest tests, 326/326 GREEN.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes brain-retro 2026-05-20 #17 — one-off Node script for investigating
the Glob p50=12.7s anomaly from initial retro. Parses transcript JSONL,
prints top-N slowest Glob round-trips with pattern + path.
Smoke-tested on session 553717ec (5h+ session): finds 32 Glob calls,
median 12690ms (matches retro finding), top-5 all 'docs/adr/**' at
20265ms — Glob recursive on ADR directory is the apparent culprit.
NOT production code path — never imported by parser/hook/analyzer.
Run on demand: node tools/glob-latency-investigator.mjs <transcript.jsonl>.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes brain-retro 2026-05-20 #16 — when the next prompt is 'neutral'
(no correction/approval/new_task markers), interpret as silent success
('no objection') and surface as soft_success. Slightly weaker than
explicit approval — labelled separately so brain-retro can show
breakdown.
4 new vitest tests, 324/324 GREEN.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes brain-retro 2026-05-20 #14 — `environment.session_turn` уже значит
'turns since last compaction' (parser counts from lastCompactIdx + 1).
Ось матрицы под именем 'session_turn' путала с глобальным turn-номером.
Семантика данных не меняется, только имя axis в FACTOR_FNS.
Existing test renamed; new explicit test verifies new name present and
legacy name absent.
1 new vitest test + 1 renamed, 320/320 GREEN.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes brain-retro 2026-05-20 #13 PIVOT — additive to F1 (parallel
session sessions session). F1 narrowed parallel_session to tool_result-only
to fix live FP. This Task adds OR-clause: Bash command containing
'git fetch && git log HEAD..origin/...' (Pravila §15.2 pre-flight)
is a strong signal that the operator expects parallel sessions.
Does NOT overwrite F1 — both signals coexist via OR.
4 new vitest tests, 319/319 GREEN.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes brain-retro 2026-05-20 #12 — each Agent tool_use produces a
subagent_invoked event with subagent_type / model (if explicit) /
first 80 chars of description. Visibility from parent Claude's
perspective; full subagent trace lives in subagents/ directory and is
out of scope for this parser.
6 new vitest tests, 315/315 GREEN.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes brain-retro 2026-05-20 #11 — parseReasoningTag extracts opt-in
<!-- reasoning: triggers="..." candidates="..." boundaries="..." -->
HTML-comment from assistant text. Semicolon-separated values merged into
heuristic-derived primary_rationale arrays via Set-dedupe.
Conservative: tag is opt-in; heuristic still runs even when tag present
(heuristic provides baseline, tag enriches).
5 new vitest tests, 309/309 GREEN.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes brain-retro 2026-05-20 #10 — STATUS.md теперь сообщает, когда
последний раз был прочитан observer (через .read-counter.json
last_read_at). Помогает не забыть про ретро между sprint-кадансами.
3 new vitest tests, 304/304 GREEN.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes brain-retro 2026-05-20 #9 — добавлены маркеры:
- correction: 'не совсем', 'другое|другая', 'не сходится', 'wrong direction'
- approval: 'класс', 'хорошо', 'принято', 'well done', 'nice'
- new_task (prefix): 'теперь', 'далее', 'следующее', 'next', 'now'
NB на JS \b с Cyrillic: \b matches word↔non-word boundary, но Cyrillic
chars не word-chars в JS RegExp default → \b после русского слова
никогда не fires. Решение: substring-match для русских correction-маркеров;
lookahead с явными разделителями для start-of-prompt new_task маркеров.
11 new vitest tests, 301/301 GREEN.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes brain-retro 2026-05-20 #8 — UserPromptSubmit hook injects
<system-reminder>...</system-reminder> blocks into user.content that
polluted classifyTask / classifyPromptSignal / routing detection.
Now stripped via regex before any analysis.
Completed by controller (Opus) after subagent hit context limit on
1250-line test file. Helper stripSystemReminders + promptText update
were committed by subagent; test cases appended via Bash heredoc.
4 new vitest tests, 290/290 GREEN.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes brain-retro 2026-05-20 #7 — each tool_result.is_error now emits
{ kind:'error', tool:<name>, summary:<first 80 chars> }. Allows
aggregation by tool (Bash/Edit/Read) + cause prefix (ENOENT/timeout/
'String to replace not found').
Required updating existing 'emits error events for tool_result with
is_error' test assertion (old shape had bare 'message' field).
4 new vitest tests + 1 existing relaxed, 286/286 GREEN.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add extractAskUserQuestionEvents() — for each AskUserQuestion toolUseResult emits
one event per question with answer_kind: option|custom|no_answer and question_count.
Integrated into parseTranscript events pipeline. 7 new tests (263 total, 0 failed).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Closes brain-retro 2026-05-20 #3 SIMPLIFIED — sanitizeWithCount in
pii-filter (counts matches per pattern) + persistent monthly counter
docs/observer/.pii-counters.json (bumped by Stop-hook on each episode
write) + status-md-generator reads real count (no more piiMatches: 0
hardcode).
PII patterns themselves NOT changed (F7 of parallel session already
extended to 13 patterns).
Counter is informational — write failure never blocks Stop-event.
5+1+1=7 new vitest tests, 256/256 GREEN.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- export extractTokenUsage(turn): sums input/output/cache/iterations/
web_search/web_fetch across all assistant messages in a turn
- parseTranscript now includes task_cost field (zero-filled when no usage)
- 7 new tests (5 unit + 2 integration); total 248/248 GREEN
- V2_FIELDS in observer-stop-hook.mjs NOT changed (backward compat)
Closes brain-retro 2026-05-20 #1 — analysis/memory-sync/regulatory-bump/
release/cleanup/monitoring/planning. Addresses '59% other' observation
from initial retro factor matrix.
Ordering: release before feature (merge feature-branch), planning before
refactor (план рефакторинга), memory-sync/regulatory-bump at top as most
specific. monitoring regex проверь состоян covers inflected forms.
9 new vitest tests, 241/241 GREEN in npm run test:tools.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
PII filter previously covered only RU phone, email, Sentry, OpenAI token,
and generic Bearer. Several common surface leaks were uncovered:
- JWT tokens (eyJ<base64>.<base64>.<base64>) — auth/session tokens.
- AWS access key IDs (AKIA<16 alphanum>) — IAM static creds.
- Yandex Cloud IAM static keys (AQVN<base64>), session tokens (t1.<base64>),
OAuth tokens (y0_<base64>) — primary cloud-provider for this project.
- IPv4 addresses (dotted-quad) — over-redacts 4-segment build numbers as
an accepted tradeoff (under-redaction is the worse failure).
- Windows user-paths (C:\Users\<name>) → C:\Users\***. Otherwise the OS
username `Administrator` leaks via task_size.files in every episode.
- POSIX /home/<name>/ → /home/***/. Same rationale for Linux dev hosts.
Pattern order: highly-specific token patterns (JWT/AWS/YC) run BEFORE
OPENAI_TOKEN/GENERIC_BEARER fallbacks; otherwise partial overlaps would
strip the wrong segments.
Tests: 9 new (each new pattern + idempotency over the expanded redaction
markers). 27/27 PII tests green.
.gitleaks.toml: added the test fixture to the path allowlist — the file
contains synthetic JWT/AWS/Yandex tokens (the filter is supposed to redact
them), not real secrets.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Bug: checkCoverage flagged anomaly when "recent commits > 0 AND episodes == 0".
Two design flaws, proven in this project:
- Wrong unit: commits = work-unit (one turn → many commits via subagent
workflow); episodes = turn-unit. A 1023-vs-19 ratio is not anomalous, it's
expected.
- Wrong window: the 14-day commit window predated the Stop-hook's existence
(registered 2026-05-19). For 13 of 14 days the hook didn't exist — 889
commits were structurally impossible to mirror as episodes.
Result: the C5 indicator was either always-red (flagging the hook's birth
as anomaly) or always-green (any episode count vs huge commit count = ok).
Either way uninformative.
Fix:
- checkCoverage(episodeCount, hookRegistered) — drops the commit param.
Warn iff hook is registered AND 0 episodes this month → the hook is
silently failing. If the hook isn't registered, 0 episodes is correct.
- runCoverageChecker derives hookRegistered from settings.json
(isObserverStopRegistered helper) and passes it to checkCoverage.
No more git execFileSync — pure fs.
Tests rewritten under the new contract: 7/7 (was 6, +1 drift-hazard guard
ensuring detail strings never mention "commit"). 15/15 coverage tests green.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
V2_FIELDS list omitted prompt_signal and events — both are always produced
by parser and buildEpisodeFromContext, so the happy path is unaffected, but
a future ctx-fallback path that dropped them would silently write a
malformed episode. Add both to V2_FIELDS; appendEpisode now throws on either
being missing.
Tests: 2 new — appendEpisode throws when prompt_signal missing /
when events missing. 38/38 stop-hook tests green.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Bug: findCausalChains flagged a chain whenever two episodes shared any
file. CLAUDE.md / MEMORY.md / STATUS.md / episodes-YYYY-MM.jsonl /
memory/*.md are touched by almost every turn (memory store, status
regeneration, normative-doc updates) — sharing them is not evidence of
causality, just baseline noise. Result: spurious chains on hot files
crowded out the genuine signal.
Fix: HOT_FILE_PATTERNS regex list + `isHotFile(path)` predicate. In
findCausalChains, filter hot files out of BOTH the errored-episode file
set AND the candidate-shared list. If only hot files were shared → no
chain. If a non-hot file is also shared → the chain stands and the
sharedFiles list contains only the non-hot ones.
Tests: 4 new cases — CLAUDE.md / memory/*.md / episodes/STATUS/MEMORY
sharing yields no chain; a turn sharing both CLAUDE.md AND /src/app.ts
yields a chain with sharedFiles=['/src/app.ts'] only. 33/33 analyzer
tests green.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Bug: inferOutcome flagged `blocked` whenever errorCount > retryCount across
the turn's events. But the parser emits an `error` event for ANY tool_result
with is_error=true — including expected failures: TDD failing-test-first,
grep returning nothing, git commands with intentional non-zero exit. On
TDD-heavy turns (project's standard discipline) this systematically marked
turns as blocked even when they ended on a successful tool_use.
Fix:
- Parser (extractProcessEvents): walk turn from end, find the LAST
tool_result; if its is_error=true, emit a single `unrecovered_error`
event. Distinguishes "turn ended on failure" from "errors recovered
later". The original per-is_error `error` events remain (useful as raw
factor signals).
- Analyzer (inferOutcome): replace `errorCount > retryCount → blocked`
with `events.some(kind === 'unrecovered_error') → blocked`. Same
ordering preserved (interrupt > blocked > rework/success/unknown).
Tests:
- Parser: emits unrecovered_error when last tool_result is_error;
does NOT emit when turn ended on a successful tool_result;
does NOT emit for turns with no tool_results.
- Analyzer: blocked iff unrecovered_error event present (not raw count);
events=[error, error, retry] → success (no unrecovered_error).
142/142 vitest green (was 128).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Bug: Claude Code's transcript JSONL file accumulates duplicated context-
rebuild snapshots — the same entry re-printed with the SAME `uuid`. Without
dedup, session_turn / task_size / events double-count, and session_turn
becomes non-monotonic across episodes parsed at different file-growth
states. Live evidence: episodes-2026-05.jsonl lines 14/15/16 of the same
session showed session_turn 139 → 140 → 91 (backwards in time). Probe
on transcript 553717ec: 22400 entries, only 6074 unique uuid (68% dup
rate); real user prompts 264 total vs 92 unique-uuid.
Fix: parseLines now tracks a `seenUuid` Set and skips entries whose uuid
has already been encountered (keep-first). Entries without `uuid`
(synthetic test fixtures) pass through unchanged. All downstream functions
(findTurnStart, extractEnvironment, extractTaskSize, etc.) operate on the
deduped entries array, so the fix is single-point and total.
Tests: new `parseTranscript — uuid-dedup` describe block covers
(1) duplicated-uuid prompts collapse → session_turn counts once,
(2) distinct-uuid entries preserved (no over-dedup),
(3) no-uuid entries pass through (synthetic-fixture safety),
(4) duplicated-uuid assistant turns → tool_calls / files_touched counted once.
110/110 parser tests green (was 106).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
extractEnvironment was scanning JSON.stringify(turn) for collision markers
(чужой staged / foreign git index / index.lock / another git process). Prose
mentions in user/assistant text flipped parallel_session=true. Live FP proven
on episodes-2026-05.jsonl line 20: my own analysis turn was non-parallel but
recorded parallel_session: true because the finding text mentioned the markers.
Fix: collectToolResultText(turn) — gather text only from tool_result blocks
(both string content and structured `[{type:text,text}]` arrays). Scan THAT
for collision markers; prose is no longer a signal.
Tests: rewrote `parallel_session narrowed` block — false on user/assistant
prose / no-tool-result turns; true on tool_result strings + structured form.
106/106 parser tests green.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
In a git worktree the shared .git/hooks/pre-commit cannot find lefthook on
PATH and silently skips every gate (pint/larastan/pest/gitleaks). This
script hardcodes the lefthook.exe + lefthook.yml paths from the main
checkout and runs `pre-commit` explicitly. Run before `git commit` inside
any worktree. Exit 0 = all gates passed.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
P0.1b: inferOutcome emits 'blocked' when a turn had more error than retry
events (an unrecovered tool failure) — previously the enum value was dead.
P0.1c: 'failure' documented as deferred to the phase-2 agent-judge. It is a
judgment (work wrong AND never corrected), not deterministically recoverable
from a transcript; a wrong-then-corrected turn surfaces as 'rework'.
P1.1: analyze() drops v1 episodes (no schema_version 2) — they lack
environment/prompt_signal/decision_provenance and polluted the factor
matrix. Reports v1SkippedCount.
P2.1: session_turn (bucketed early/mid/late) and parallel_session added to
FACTOR_FNS — closes the schema↔matrix mismatch (both were captured in the
episode but absent from the factor axes).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
P0.2: count session_turn from the last compaction. The transcript file
accumulates duplicated context-rebuild snapshots (quirk #101), so counting
real prompts from i=0 inflated it and made it non-monotonic. Now counts
"real prompts since the last compaction" — monotonic by construction.
P0.1a: widen the correction prompt_signal regex (не работает / сломал /
опять / откати / revert / still not / wrong / ...). The old regex was too
narrow, so rework outcomes were invisible to the factor analysis.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Browser smoke (Playwright) revealed that rewriting path internally without
changing the response URL left the browser's base URL as /, breaking
relative <script src="dashboard.js"> and ../automation-graph-data.js
references. 302 redirect makes the browser settle on /docs/observer/,
which resolves the relative paths correctly. All 4 views verified clean
(0 console errors). Screenshots: brain-dashboard-{map,replay,feed,aggregate}-view.png.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Worktree has no app/node_modules — vitest not run here; final regression
deferred to main-checkout post parallel-session release. Logic is a 7-line
pure filter; tests cover empty filter, classification, errors-only.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
#1 — detectAskUserQuestionChoice: when a turn contains an AskUserQuestion
whose answer exactly matches an offered option label, classify as
user_chose_from_options. The answered entry carries a structured
toolUseResult (questions[].options[].label + answers map). A custom
"Other" free-text answer is NOT a pick — falls through. Wired into
parseTranscript after the text-list detector.
#3 — parallel_session: dropped broad word matches (параллельн /
"parallel session") that false-fired on any casual mention. Now only
strong collision evidence (foreign git index / чужой staged /
index.lock / another git process). Best-effort per spec R2 — prefer
false-negative over false-positive.
169/169 tools tests GREEN (+9 new).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Root cause (systematic-debugging): isRealUserPrompt treated skill-content
("Base directory for this skill:"), local-command output
(<local-command-stdout>), and interrupt markers as genuine prompts.
findTurnStart then anchored a turn on the synthetic message — the turn
slice missed the genuine prompt's UserPromptSubmit hook_additional_context
attachment → economy_level: null, wrong prompt_signal/task_classification.
Same cause made extractLastUserPromptText return skill content, so the
Stop-hook routing-gate false-positive-blocked autonomous §12 skill
invocations (detectMethodDirected saw the node name in skill text).
Fix: SYNTHETIC_PROMPT_MARKERS + isSyntheticPrompt — isRealUserPrompt
returns false for synthetic messages. One fix closes both the
economy_level capture gap and the 2nd routing-gate FP class.
160/160 tools tests GREEN (+3 new).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
buildFactorMatrix already buckets decision_provenance.kind dynamically
(brain-retro-analyzer.mjs:112) — no production change needed. Test
pins that user_chose_from_options is counted on the provenance axis.
12/12 brain-retro tests GREEN.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
When episode is user_chose_from_options, routing-gate does NOT block —
collaborative-choice from Claude-offered options doesn't require a
routing-tag (detector is deterministic). 18/18 stop-hook tests GREEN.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
detectChoiceProvenance runs BEFORE parseRoutingTag; if last assistant
turn offered options and user prompt references one, decision_provenance
becomes user_chose_from_options. Otherwise falls back to existing
routing-tag / autonomous logic.
3 new parser tests GREEN; all existing tests still GREEN (43/43).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Pure module — extracts options (numbered/lettered/bullets/AskUserQuestion)
from last assistant message, detects user reference (position-based +
substring), returns decision_provenance for the 3rd kind.
23/23 tests GREEN.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Pure deterministic Layer-4 aggregation module (spec §6) for the /brain-retro
skill. Exports: dedupeEpisodes, inferOutcome, groupEpisodesToTasks,
findCausalChains, buildFactorMatrix, analyze. Read-only — never writes JSONL.
11/11 tests green. CLI smoke: 10 real episodes → valid JSON with all 5 keys.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Дмитрий