菜单
Break work into small verifiable tasks with acceptance criteria and dependency ordering
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
Use when an AI tool gains support for a feature agentsmesh does not yet expose — or exposes wrongly — and you must raise or correct an existing target's capability for project and/or global scope (rules, additionalRules, commands, agents, skills, mcp, hooks, ignore, permissions). Also use to audit every target for capability gaps, or to act on an audit/feature-request someone else wrote. Start by verifying the claim against the tool's OWN primary docs — such findings are frequently fabricated, stale, or wrong-scoped, so be ready to reject. Covers verification-first research, descriptor wiring, round-trip symmetry, global-only scopeExtras emission, settings-merge and plugin safety, format/path-change blast radius, strict TDD, two-stage review, and matrix/docs sync. Start with `pnpm capabilities:audit`; research only STALE/MISSING cells; the ledger memoizes verified formats so audits stay fast.
Use when an AI tool gains support for a feature agentsmesh does not yet expose — or exposes wrongly — and you must raise or correct an existing target's capability for project and/or global scope (rules, additionalRules, commands, agents, skills, mcp, hooks, ignore, permissions). Also use to audit every target for capability gaps, or to act on an audit/feature-request someone else wrote. Start by verifying the claim against the tool's OWN primary docs — such findings are frequently fabricated, stale, or wrong-scoped, so be ready to reject. Covers verification-first research, descriptor wiring, round-trip symmetry, global-only scopeExtras emission, settings-merge and plugin safety, format/path-change blast radius, strict TDD, two-stage review, and matrix/docs sync. Start with `pnpm capabilities:audit`; research only STALE/MISSING cells; the ledger memoizes verified formats so audits stay fast.
Use when about to edit a file or run a state-changing command, or after any failure, correction, or surprising result.
Use when about to edit a file or run a state-changing command, or after any failure, correction, or surprising result.
Use when about to edit a file or run a state-changing command, or after any failure, correction, or surprising result.
A relentless interview to sharpen a plan or design.
| name | am-command-planning |
| description | Break work into small verifiable tasks with acceptance criteria and dependency ordering |
| x-agentsmesh-kind | command |
| x-agentsmesh-name | planning |
Invoke the planning-and-task-breakdown skill.
Read the existing spec (SPEC.md or equivalent) and the relevant codebase sections. Then:
Save the plan to tasks/plan.md and task list to tasks/todo.md.
