菜单
Runs configurable multi-round research with compare, review, and explore modes, visible progress, source collection, and final sourced answers.
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
Runtime instructions for Ask mode: answer directly without code mutation.
Creates and maintains project-board tasks with correct dedupe, umbrella/root linking, dependencies, tags, authors, assignments, and autopilot queue placement.
Runtime instructions for Plan mode: produce implementation or investigation plans without code mutation.
Ultra-compressed communication mode. Cuts token usage ~75% by speaking like caveman while keeping full technical accuracy. Supports intensity levels: lite, full (default), ultra, wenyan-lite, wenyan-full, wenyan-ultra. Use when user says "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", or invokes /caveman. Also auto-triggers when token efficiency is requested.
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me".
Test-driven development with red-green-refactor loop. Use when user wants to build features or fix bugs using TDD, mentions "red-green-refactor", wants integration tests, or asks for test-first development.
| name | deep-research |
| description | Runs configurable multi-round research with compare, review, and explore modes, visible progress, source collection, and final sourced answers. |
| userInvocable | true |
| allowedTools | ["web.search","web.fetch","planning.progress_update","runtime.exec"] |
Use this skill when the user asks for /deepresearch or explicitly requests deep research.
Configuration is supplied as:
mode: compare, review, or explorerounds: integer from 1 to 8Treat each round as one search plus synthesis cycle:
web.searchweb.fetchplanning.progress_updateMode behavior:
compare: compare alternatives, criteria, tradeoffs, and evidence.review: evaluate one subject, claim, product, document, or proposal with strengths, risks, and verdict.explore: map a topic, identify themes, unknowns, and promising follow-up directions.The final answer must include:
Do not decide progress, completion, or source quality by matching assistant prose. Use tool results, fetched source content, and explicit progress updates.