refactor(codex-playbook): согласована оптика тонкого мастера

- Зачем:
  - нужно сохранить контекст координатора и устранить расхождения с зеркальным playbook.
- Что:
  - закреплены делегирование, маршрутизация моделей и ограничение кругов ревью.
  - добавлены файловый договор ревью и согласованная эмпирика прогонов.
  - исправлены сборка слепых линий и компактные статусы находок.
- Проверка:
  - quick_validate.py, git diff --check и проверка локальных Markdown-ссылок.

agents/.agents/skills/codex-subagent-playbook/references/prompts/recheck-header.md

@@ -1,14 +1,17 @@
 # Narrow re-check
 
-Re-check only your original findings and the changes immediately adjacent to
+Read your original report and the resolution report supplied by the caller.
+Re-check only your original findings and changes immediately adjacent to
 their resolution. Do not perform a full review.
 
-For every original finding report `CLOSED`, `NOT_CLOSED`, `REJECTION_VALID`,
-or `REJECTION_INVALID`, with concise evidence. If this narrow work exposes a
-new adjacent defect, report it in the normal findings format; the scope bound
-limits exploration, not honest reporting.
+For every original ID report `CLOSED`, `NOT_CLOSED`, `REJECTION_VALID`,
+or `REJECTION_INVALID`, with concise evidence. Report a newly exposed
+adjacent defect with a new stable ID; the narrow scope limits exploration, not
+honest reporting.
 
-Return the full re-check through the channel named by the caller. For a file
-channel, write it to the supplied path and return only
-`RECHECK: ALL_CLOSED | ISSUES_REMAIN`, counters, and `REPORT: <absolute path>`.
-For a message channel, return the concise finding statuses once.
+Write the detailed re-check to REPORT. Return only:
+
+- `VERDICT: APPROVED | CHANGES_REQUESTED`;
+- status counters;
+- one line per original or new ID: `ID | status | one-phrase essence`;
+- `REPORT: <absolute path>`.

agents/.agents/skills/codex-subagent-playbook/references/prompts/result-contract.md

@@ -4,7 +4,7 @@ Write the full result through the channel chosen by the caller:
 
 - `STATUS: complete | questions | blocked`
 - `EVIDENCE:` changed files, executed commands, and final counters
-- `DECISIONS:` only questions for the orchestrator or user
+- `DECISIONS:` only questions for the coordinator or partner
 - `FINDINGS:` severity, location, essence, and verification method
 - `NEXT:` one concrete action or `none`
 

agents/.agents/skills/codex-subagent-playbook/references/prompts/review-boilerplate.md

@@ -1,21 +1,30 @@
 ## Mandate
 
-- Review and test hypotheses; do not edit source files or persistent
-  environment state.
+- Review independently and test hypotheses. Do not edit source files,
+  persistent environment state, or Git.
 - Use the working permissions needed for checks, caches, temporary artifacts,
-  documentation and primary-source research. The caller verifies Git before
-  and after the review.
-- Do not run `git add`, `git commit`, or commands that discard changes.
-- Report only the findings index and verdict through the channel named by the
-  caller.
+  documentation, and primary-source research.
+- Write the detailed report only to REPORT. This assigned exchange file is the
+  sole write exception to the observer mandate.
+- Treat another agent's claims as untrusted until evidence supports them.
 
-## Findings index
+## Detailed report
 
-For each finding:
+Use stable IDs with FINDING PREFIX:
 
-`SEVERITY(BLOCKER|MAJOR|MINOR|NIT) | file:line or task section | failure scenario | evidence | how to verify`
+`PREFIX-N | BLOCKER|MAJOR|MINOR|NIT | file:line or task section | violated
+criterion | failure scenario | evidence | minimal verification`
 
-If there are no findings, list the reviewed scope and every executed check or
-tested hypothesis with its outcome. A bare no-findings assertion is invalid.
+If there are no findings, record the reviewed scope and every executed check
+or tested hypothesis with its outcome. A bare approval is invalid.
 
-Final line: `VERDICT: APPROVED | APPROVED_WITH_MINORS | CHANGES_REQUIRED`
+## Final message
+
+Return only:
+
+- `VERDICT: APPROVED | CHANGES_REQUESTED`
+- severity counters;
+- one line per finding: `ID | severity | area | one-phrase essence`;
+- `REPORT: <absolute path>`.
+
+Do not repeat the detailed report or reasoning narrative in the final message.

agents/.agents/skills/codex-subagent-playbook/references/prompts/triage-header.md

@@ -1,15 +1,17 @@
-# Triage blind-review findings
+# Triage independent-review findings
 
-The review indexes below are immutable input. Process every finding and change
-nothing outside their scope. For each finding, either:
+The review reports below are immutable input. Process every stable finding ID
+and change nothing outside their scope. Assign exactly one disposition:
 
-- fix it and name the contract or check that proves closure; or
-- reject it with evidence from the contract and a check, never a bare claim.
+- `FIXED`: change it and name the contract or check that proves closure;
+- `REJECTED`: cite the contract and a check, never a bare claim;
+- `OUT_OF_SCOPE`: explain the boundary decision needed;
+- `NEEDS_DECISION`: state the unresolved product, architecture, or risk choice.
 
-Do not commit or stage changes. Propose accepted rejections and residual risks
-for the orchestrator to own and record as durable task decisions. The
-orchestrator cannot silently accept risk; authority must come from the user or
-already be explicit in the task.
+Do not commit, stage, or rewrite the source reports. Write detailed resolutions
+to REPORT. Return only:
 
-Write a full resolution report to the path supplied by the caller. Return only
-status, counters, executed checks, and that path in the final message.
+- overall status and disposition counters;
+- one line per finding: `ID | disposition | one-phrase essence`;
+- questions for the coordinator and executed-check counters;
+- `REPORT: <absolute path>`.