菜单
Durable professor-critic research workflow bound to a tracked goal, with validation-gated completion. Use when a research mission needs ongoing goal-focused management plus professor/critic-style validation.
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
Run a regression-tests-first, smell-by-smell deslop/refactor workflow that preserves behavior and raises signal quality. Use when the user asks to "cleanup", "refactor", "deslop", or when follow-up code has duplicates, dead code, weak boundaries, or fallback-like slop.
Run read-only deep repository analysis and return a ranked synthesis with explicit confidence, concrete file references, and clear evidence-vs-inference boundaries. Use when a user says "analyze", "investigate", "why does", "what's causing", or needs grounded cross-file explanation before any changes are proposed.
Deprecated compatibility shim for Claude advisor requests. Use `ask` with the claude backend instead.
Deprecated compatibility shim for Gemini advisor requests. Use `ask` with the gemini backend instead.
Ask a locally installed external advisor CLI (Claude or Gemini) for focused questions, reviews, brainstorming, or second opinions, and capture a reusable artifact. Use when the user wants a Claude/Gemini second opinion, "ask claude", or "ask gemini".
Strict autonomous delivery loop that runs deep-interview, ralplan, ultragoal, code-review, and ultraqa in order with automatic re-planning when gates are not clean. Use on "autopilot", "autonomous", "build me", "create me", "make me", "full auto", "handle it all", or "I want a/an…".
| name | autoresearch-goal |
| category | research |
| status | active |
| description | Durable professor-critic research workflow bound to a tracked goal, with validation-gated completion. Use when a research mission needs ongoing goal-focused management plus professor/critic-style validation. |
| agent_created | true |
| triggers | ["autoresearch-goal","research goal","professor critic research","validated research mission"] |
Ported from oh-my-codex
autoresearch-goal. OMX runtime conventions ($macroinvocation,omxCLI,.omx/state directory) are replaced with WorkBuddy idioms (Skill tool, Agent tool, task list,.workbuddy/memory).
Use this workflow when a research mission should be bound to a tracked goal while WorkBuddy remains the durable state owner. This is for research projects that need goal-focused management plus professor/critic-style validation; it is not the default answer for ordinary pre-planning best-practice lookup.
.omw/autoresearch-goal/<slug>/.autoresearch-goal writes:
.omw/autoresearch-goal/<slug>/mission.json.omw/autoresearch-goal/<slug>/rubric.md.omw/autoresearch-goal/<slug>/ledger.jsonl.omw/autoresearch-goal/<slug>/completion.json.omw/autoresearch-goal/<slug>/mission.json and rubric.md (topic, rubric, and the critic command/prompt)..omw/autoresearch-goal/<slug>/ledger.jsonl via the verdict command (verdict pass|fail|blocked + evidence).verdict=pass. After the mission audit passes, mark the task-list goal complete and record .omw/autoresearch-goal/<slug>/completion.json with the matching goal snapshot.A passing professor-critic artifact and a matching completed goal snapshot (task-list status complete) are required. Assistant prose, partial tests, or a failed/blocked verdict are not sufficient.
Lifecycle: the task-list goal starts active when the mission is created, and is marked complete after the professor-critic and audit pass. Local artifacts reconcile snapshots and record the cleanup step; they must not mutate hidden external goal state.