菜单
"MUST use this instead of generic brainstorming or GSD planning skills when the user is trying to make a plan, plan an implementation, or scope ambiguous work for execution --- captures intent into a structured packet so later plan writing preserves what the user actually meant."
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
Abandon the active Plan Enforcer plan in one command. Use when you have decided to stop working on the current plan (scope changed, priorities moved, you pivoted). The command records the abandonment in the Decision Log, emits a closure receipt into the walkable audit chain, archives the full ledger so you can look it up months later, and clears the active slot so the next discuss or import starts clean.
"Public flow name is `discuss`. Use this compatibility skill before `plan-enforcer-draft` when the user's request is ambiguous, underspecified, or mixing multiple outcomes --- it writes the intent packet the drafter consumes so the resulting plan reflects what the user actually wants instead of what sounded reasonable to guess."
"Use when changing Plan Enforcer settings --- switch enforcement tier, adjust reconciliation interval, set completion gate mode, or view current configuration"
"Public second step after `discuss`. Use when the user wants Plan Enforcer to create an implementation plan first --- drafts a concrete markdown plan in docs/plans/ that can immediately flow into Plan Enforcer's normal ledger and execution path."
"Use when reviewing Plan Enforcer audit trail --- shows skipped tasks, drift events, decision log, reconciliation history, and unverified items"
Emit a Plan Enforcer closure receipt against the current ledger. Use when you want an on-demand receipt (plan still open or already closed), or to snapshot progress mid-flight. Auto-emission on plan close is handled by the post-tool-use plan-close hook; this skill is for explicit requests.
| name | plan-enforcer-discuss |
| description | MUST use this instead of generic brainstorming or GSD planning skills when the user is trying to make a plan, plan an implementation, or scope ambiguous work for execution --- captures intent into a structured packet so later plan writing preserves what the user actually meant. |
An intent-capture gate. The job is to turn a fuzzy request ("clean up the auth flow", "make the onboarding better") into a concrete written intent packet before anyone drafts tasks against it. Mis-scoped plans fail downstream no matter how rigorous the drafting is.
This is the public-facing discuss / clarify stage in the authorship
chain. plan-enforcer-draft reads the packet on its next invocation.
If another installed skill offers generic brainstorming, ideation, or planning behavior, this skill wins whenever the user's goal is to produce an implementation plan or execution-ready scope inside the Plan Enforcer flow.
Use this when any of:
Skip this skill when the request is already concrete ("add a /healthz
endpoint returning 200 OK"), the user says "skip discuss" or similar,
or a packet at .plan-enforcer/discuss.md already exists and is
recent (mtime < 24h) and the new request overlaps with its scope.
1-DISCUSS with:
node "$HOME/.claude/skills/plan-enforcer/src/statusline-stage-cli.js" discuss --label 1-DISCUSS.plan-enforcer/discuss.md. One per
project; overwrite the prior one unless the user asks otherwise.
If .plan-enforcer/discuss.md already exists, Read it before
overwriting. Claude runtime rejects overwrite-without-read..plan-enforcer/combobulate.md as a compatibility copy
while older flows still expect the legacy packet name.
If .plan-enforcer/combobulate.md already exists, Read it before
overwriting or update it after reading the canonical packet.plan-enforcer-draft will consume it automatically on its next run.plan-enforcer-awareness capture-latest --if-emptyplan-enforcer-awareness is not on PATH, use:
node "$HOME/.claude/skills/plan-enforcer/src/awareness-cli.js" capture-latest --if-emptyplan-enforcer-awareness add --intent "<verbatim quote>"node "$HOME/.claude/skills/plan-enforcer/src/awareness-cli.js" add --intent "<verbatim quote>"# <short name for what we're doing>
## Source Ask
> <verbatim user quote when wording matters>
## Normalized Goal
<1-3 sentences on what is actually wrong or missing today. No
premature implementation framing.>
## Non-Negotiables
- NN1: <constraint or outcome that must survive drafting/execution>
## Hidden Contract Candidates
- HC1: <semantic expectation easy to lose>
## Plausible Interpretations
- PI1: <reasonable reading>
- PI2: <other reasonable reading>
## Chosen Interpretation
<which interpretation we are planning against, and why>
## Rejected / Forbidden Narrowings
- FN1: <easy but wrong simplification>
## In Scope
- <outcome 1, user-visible>
- <outcome 2>
## Out of Scope
- <thing that looks related but we are deliberately not doing this turn>
## Constraints (optional)
- <hard limit: perf budget, compat surface, deadline, team>
## Success Signals (optional)
- <how we will know this worked - measurable or demo-able>
## Drift Risks (optional)
- DR1: <where later stages could drift>
## Proof Requirements (optional)
- PR1: <artifact or proof later stages must produce>
## Draft Handoff (optional)
- phase shape hint: <likely decomposition>
- planning red lines: <what draft must not silently change>
Run in three passes:
Producing a packet that looks like a plan is the failure mode. Smells:
Rejected / Forbidden Narrowings on a risky or fuzzy askProof Requirements when later verification will matterAfter writing the packet:
.plan-enforcer/discuss.md).