菜单
dure audit — run the deterministic repo scanner and surface findings (debt / missing-test / progress-consistency signals) as evidence for /dure:interview and /dure:plan. Report-only; never creates issues.
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
dure direction — gather survey/audit/status evidence, run a deep interview that converges a project direction, write a validated .dure/directions/<slug>.md proposal, and hand the chosen direction to /dure:plan. Report-only at the validator boundary; never writes the roadmap.
dure survey — run the deterministic state survey and surface forward-looking planning signals (closable milestones, empty epics) as evidence for /dure:interview and /dure:plan. Report-only; never creates issues.
dure deep interview — interview a vague requirement until understanding converges quantitatively, then crystallize it into .dure/specs/<slug>.md. Four traits — evidence-grounded, red-team, auto-research, quantitative convergence gate.
dure sync — idempotently mirror the local roadmap to GitHub Issues/Milestones. plan/apply separation; structure pushed, status pulled, conflicts reported. gh CLI first.
dure progress report — aggregate per-item roadmap status into per-milestone completion, blockers, and local↔GitHub conflicts.
dure decomposition — decompose the active fixed spec into milestone/epic/issue per-item files (source of truth), review them, and regenerate the index. Every issue gets a non-empty, testable acceptance criterion.
| name | audit |
| description | dure audit — run the deterministic repo scanner and surface findings (debt / missing-test / progress-consistency signals) as evidence for /dure:interview and /dure:plan. Report-only; never creates issues. |
| argument-hint | |
| allowed-tools | Bash Read |
| disable-model-invocation | true |
Goal: per specs/dure-audit.md, run the deterministic audit and present its findings as evidence for planning. The interview/report runs in the user's language; artifacts stay English (ADR-0007).
.dure/roadmap/ and MUST NOT create issues.
Findings are evidence; the user (or /dure:interview / /dure:plan) decides what becomes work.dure-doctor: doctor owns integrity (schema/hierarchy/orphans); audit owns
debt / missing-test / progress-consistency signals.${CLAUDE_PLUGIN_ROOT}/scripts/dure-context.sh to confirm the target repo.${CLAUDE_PLUGIN_ROOT}/scripts/dure-audit.py
→ JSON {status, counts, findings:[{check, severity, file?, line?, id?, message}]}.check then severity
(error > warning > info), each with its file:line or id. Lead with the counts summary.
If findings is empty, say so plainly./dure:interview "<finding>" as grounding evidence,${CLAUDE_PLUGIN_ROOT}/scripts/dure-audit.py --suggest to get draft issue stubs
(title + suggested acceptance) for warning-or-higher findings, which the user can hand to /dure:plan, or/dure:plan.
Let the user choose; do not create issues or edit the roadmap.Implementation status: wraps the tested
scripts/dure-audit.py(3 checks: todo-marker, untested-script, done-parent-undone-child). Auto-running onSessionStartor feeding findings straight into/dure:planare possible later increments, kept out of this report-only slice.