菜单
Diff/PR review: council → thermos/yagni → delta BPR → Sol merge. Report only.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Plan a huge chunk of work — more than one agent session can hold — as a shared map of decision tickets on your issue tracker, and resolve them one at a time until the way is clear. On ticket close, may offer ADRs (promotion into docs/adr/) when ADR-POLICY criteria hold.
Sharpen a plan or design against this codebase via grilling + domain-modeling (CONTEXT.md / ADR offers). May offer /wayfinder when fog appears. Invoke with `/grill-with-docs`.
Applies Dockerfile, Docker Compose, BuildKit, and container security patterns for local development and hardened deployable images. Use when authoring Dockerfiles or compose files, wiring secrets (runtime vs build-time), reproducible bases, PID 1 and healthchecks, volumes and networking, supply-chain hygiene, or troubleshooting compose stacks.
Routes Go work to the right reference guides and conventions for the task. Use when working on Go architecture, implementation, refactoring, concurrency, error handling, testing, performance, or code review.
PostgreSQL database patterns for query optimization, schema design, indexing, and security. Use when writing SQL or migrations, designing schemas, troubleshooting slow queries, implementing Row Level Security, tuning connection pooling, or reviewing database changes. Based on Supabase-leaning Postgres practice.
Applies Python idioms, PEP 8 norms, typing, packaging, concurrency, tooling, and performance discipline to everyday Python code. Use when writing, reviewing, refactoring, or packaging Python; when the user names type hints, dataclasses, pytest, ruff/black/mypy, asyncio, concurrency, EAFP vs LBYL, or Python idioms; or asks for a Python-specific code-quality pass.
| name | code-review |
| description | Diff/PR review: council → thermos/yagni → delta BPR → Sol merge. Report only. |
Post-change code review. Report only — do not edit code unless the user asks after the report.
Probe Task model / subagent_type before fan-out. Route per ../council/references/cursor-task-workflow.md, with these overrides:
| Role | Model |
|---|---|
| Council workers | Workflow defaults (Composer lane for partitions) |
Thermos, /yagni | Grok lane |
| Fresh pass + merge | gpt-5.6-sol-medium if in the enum; else Grok lane |
Delta /best-practices-research runs as that skill is written (it owns its Task fan-out and model picks).
Scope. Honor the user's prompt. If unspecified: changes vs main including working-tree WIP (committed on the branch + staged + unstaged). Done when: base, WIP inclusion, and any path/PR narrowing are explicit.
Package. Build the review package: git diff for that scope + full contents of every changed file (shell + explore Tasks as needed). Stop in one line if the diff is empty. Done when: the package holds diff output and every changed file's contents, or an empty-diff stop.
Council. Run /council scoped only to areas the change touches — context brief for specialists, not the review itself. Attach that brief to later Task prompts. Done when: every touched area has been explored enough to brief steps 4–6.
Specialists. In one message, two parallel Grok-lane Tasks (each gets the package + council brief):
subagent_type matching thermos review in the enum (thermo-nuclear-review-subagent preferred). Prompt with ### Git / diff output and ### Changed file contents. If no thermos type exists, say so and stop (Thermos plugin required).generalPurpose (readonly). Run the /yagni skill (read ../yagni/SKILL.md if not already loaded); review the change set; return findings only.
Done when: both have returned.Delta BPR. Run /best-practices-research only when the diff adds a new domain/library/API surface not covered by an earlier recon artifact, or thermos/yagni flags uncertain conventions. Reuse prior recon; research only what's new. Otherwise skip with a one-line reason. Done when: skipped with reason, or delta findings are in the package.
Fresh pass + merge. One Task (generalPurpose or a review/judge type if in the enum; readonly). Prompt must include the package, council brief, and full thermos / yagni / BPR outputs. Worker: (1) independent fresh pass over the package, (2) merge all lenses — dedupe, calibrate severity, attribute each finding's lens. Model per the table above. Done when: a single prioritized report is ready.
Report. Give the user the merged report (shape). Do not fix. Done when: every finding has severity, file:line, summary, and lens; or empty diff was already stated.
Do not nest council, BPR, or yagni inside the Sol Task. Do not run full BPR by default.