Дмитрий
b632bcbae6
V2 audit found 10 new bypasses: - Fatal: subagent inheritance via text-prefix (no enforcement) - Critical: hook race conditions / DoS timeout / Bash script execution - Serious: path-deny symlinks/case / AskUser fatigue / silence off-topic - Edge cases: parallel subagents / coverage-verify interaction / sub-shells V3 closes all 10: - 3.2 rewritten: env-based subagent inheritance (env vars + inheritance file), gate-on-subagent reads parent state via CLAUDE_GATE_INHERIT env - 3.4 new: subagent constraints (no AskUser, no recursive Task, max 3 parallel) - 3.5 new: atomic writes (tmp+rename) + proper-lockfile for race-free state - 3.6 new: gate budget 2s + state cache TTL 5s + lazy transcript parsing - 3.1 extended: path normalization (resolve + realpath + case-fold + env) - 4.5 extended: max 2 AskUserQuestion per turn (fatigue exploit closed) - 4.7 extended: off-topic at silence uses task_classification - 5.1 extended: file-watcher for script execution + broad sweep sub-shell blacklist (backticks, command sub, process sub, heredocs) - 5.2 new: static content scan for node/python/vitest scripts before exec - 7.1 new: coverage-hint coordination layer between gate and coverage-verify Implementation cost: 8.5-12h (v2) to 13.5-20h (v3). +5-8h for architectural fixes (env-inheritance, atomic writes, static scan) + infrastructure (gate budget, path norm, askuser counter, coverage-hint). V2 baseline preserved as commitb510a758. V1 as7a43c175. cspell-words.txt += shutil / rmtree. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Observer infrastructure
Passive evidence-loop for the Лидерра «brain» per ADR-011.
Files
episodes-YYYY-MM.jsonl— append-only JSONL, one line per Stop-event. Schema v2 (schema_version: 2): the 5 mandatory fields +decision_provenance(who chose the node),environment(economy_level / model / post_compaction / session_turn / parallel_session),task_size,task_ref,prompt_signal, and anoutcomethat isunknownat write time (refined by/brain-retro). On an internal hook failure a minimalobserver_errormarker line is written instead of a silent skip. Written bytools/observer-stop-hook.mjsviatools/observer-transcript-parser.mjs.notes/YYYY-MM-DD-<slug>.md— optional MD notes for sessions with qualitative history.STATUS.md— auto-generated dashboard. Regenerated per-commit bytools/status-md-generator.mjs..read-counter.json— C3 observer-of-observer counter. Updated on Read of observer files.dashboard.html+dashboard.js+dashboard-core.js— Brain Dashboard: visualises the episode log over the automation-graph topology (4 views — Карта / Разбор / Лента / Агрегат). Runnpm run brain:dashboard, open the printed localhost URL.dashboard-core.jsis pure logic, unit-tested intools/brain-dashboard-core.test.mjs.
Lifecycle
- Write: every Stop-event appends one JSONL line, parsed from the session transcript (Stop-hook).
- Aggregate:
/brain-retroskill reads JSONL each sprint, proposes regulatory candidates. - Surface:
STATUS.mdshows controllers + monthly stats. - Self-prune: C3 warns if 54 weeks pass without any read of observer files.
Routing-tag discipline
When the user dictates a specific method/node (e.g. «запусти discovery-interview»), Claude must emit one line in its response:
<!-- routing: provenance=user_directed_method node=<chosen> counterfactual=<node Claude would have chosen autonomously> -->
The Stop-hook routing-gate (tools/observer-routing-detector.mjs + routingGateDecision) detects a dictated method; if the tag is missing it returns decision: block, so the turn cannot end without the tag. The gate fires at most once per turn (stop_hook_active guard). This makes decision_provenance reliable — factor analysis can separate a router error from a user-dictated one.
Privacy
PII filter (phone numbers, emails, tokens) is applied before every write — see tools/observer-pii-filter.mjs. gitleaks pre-push also scans observer files as part of full-history sweep.
Don't
- Don't edit
episodes-*.jsonlmanually — it's append-only. - Don't write outside
docs/observer/notes/for hand-curated notes. - Don't change
.read-counter.jsonmanually — it's maintained by hooks.
HK1 pre-check (Pravila ADR-010) — verified 2026-05-19
Before registering tools/observer-stop-hook.mjs on Stop event (Task B5), verified collision against 6-component economy/skill-discipline architecture:
- User-level
~/.claude/settings.jsonalready has Stop hook: agent-type Sonnet-4.6 economy compliance verifier (analyzes transcript for claim-without-evidence violations). - Project-level
.claude/settings.json— Stop slot empty.
Result: no overwrite. observer-stop-hook will be added as command-type entry in project-level Stop array. Project + user scopes are independent slots in Claude Code 2.x — both run on the same Stop event without conflict. The agent verifier (user scope) and the JSONL appender (project scope) have non-overlapping responsibilities.