菜单
Explicit-request macro. Activate only when the user directly requests this macro; never infer activation from task characteristics. Skill — Run N blind agents on the same job in parallel, then merge findings into consensus/unique/conflicts. Agents report only — no edits for concurrent safety.
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
User-invoked shorthand that inverts delegation — produce a filesystem bundle (checklist, drop-bin directory, notes template) of human-only actions and artifacts blocking project progress, and launch the user's editor on it.
Explicit-request macro. Activate only when the user directly requests this macro; never infer activation from task characteristics. Skill — Shorthand that tells the agent to ask clarifying questions instead of making decisions on its own
Explicit-request macro. Activate only when the user directly requests this macro; never infer activation from task characteristics. Skill — Sequence a task as ordered parts that add up to the whole. Each iteration may read prior iterations' outputs.
Explicit-request macro. Activate only when the user directly requests this macro; never infer activation from task characteristics. Skill — Use subagents to save context space and/or parallelize subtasks where possible
Explicit-request macro. Activate only when the user directly requests this macro; never infer activation from task characteristics. Skill — Spawns an independent subagent to critique recent work with web research. Use when the user wants a second opinion, says "doubt this", or wants to verify code against docs.
Explicit-request macro. Activate only when the user directly requests this macro; never infer activation from task characteristics. Skill — Search prior coding-agent chat transcripts (Claude Code, Codex, ...) for context. Use when the user explicitly activates this skill or asks to recall or dig up something from past chats.
| name | consensus |
| description | Explicit-request macro. Activate only when the user directly requests this macro; never infer activation from task characteristics. Skill — Run N blind agents on the same job in parallel, then merge findings into consensus/unique/conflicts. Agents report only — no edits for concurrent safety. |
Honor every skill explicitly activated in the user's request exactly once. If another activated skill is not yet loaded and the host provides a skill-loading mechanism, load it through that mechanism. Do not reload an active skill.
Do NOT re-invoke this skill recursively. Do NOT re-read these instructions or any other document in a loop. If you encounter any error or are unsure how to proceed, STOP and tell the user. Execute the workflow below once, then stop.
You are dispatching N independent, blind agents to work on the same job in parallel, then merging their findings.
Parse the user's request for a number:
2, 3, 4, 5) → agent count. Cap at 5.The job itself comes from the conversation context or from a composed skill.
Identify what the agents should work on:
macros:doubt), that skill defines the job and prompt template.Prepare the agent prompt from the job. Prepend this constraint to each agent's prompt:
You are one of ${N} independent agents reviewing the same work.
Report your findings but do NOT apply fixes.
List what you would fix and how. Do not modify any files.
The no-edits constraint is non-negotiable. Parallel agents writing to the same files will produce conflicts and lost work.
Launch all N agents concurrently, batching the dispatch if the host supports it. Each gets the identical prompt. They are blind to each other.
After all agents return, merge their findings:
Present one unified, terse list to the user. No per-agent sections.