Дмитрий
32cdf352d0
Adversarial review of v1 found 10 bypass paths:
- Fatal: AskUserQuestion = universal unlock (1 question = full bypass)
- Critical: Bash unlimited / Skill-invoke-no-followthrough / State-files edit
- Serious: Subagent inheritance / Direct-invocation regex too wide
- Edge cases: leading questions / multi-direct / failure mode / chain TTL
V2 closes all 10:
- 3.1 Protected paths (hard-deny for runtime/settings/skill/hook files)
- 3.2 Subagent gate inheritance via subagent-prompt-prefix injection
- 3.3 Failure modes — explicit fail-CLOSE policy
- 3 chain-state TTL 24h + explicit-clear markers
- 4 Direct invocation = strict whitelist (slash-cmd / Skill() / used N / делай exact)
- 4 Multiple direct invocations require AskUser between executions
- 4.5 AskUserQuestion answer parsing — gate reads transcript response,
unlocks only for explicitly-approved action, blocks on stop answer
- 4.6 Post-skill partial unlock (Read/Grep/next-chain-step allowed,
Edit/Write require additional AskUser, Bash requires whitelist+approval)
- 4.7 Question quality detector in rationalization-audit (blocks
missing-stop-option, flags leading options, off-topic questions)
- 5.1 Bash content rules: whitelist read-only / hard-blacklist mutating /
conditional-whitelist after AskUser approval / path-deny overlay
Implementation cost: 6-8.5h (v1) to 8.5-12h (v2). +2.5-3.5h for
Bash content parser, answer parser, question-quality detector,
hard-deny logic.
Spec v1 (commit ae5ea663) remains in git as baseline.
cspell-words.txt += детектирован / fgrep / chgrp.
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.