菜单
Public demo for Docs Drift: a feature changes but documentation is stale, and the drift state is surfaced before a second doc fix. Use when a user asks to run, preview, or collect evidence for docs drift with the current Claude Code or Codex observer session.
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
Single Aming Claw skill entrypoint. Use this first for Aming Claw governance, backlog, graph, runtime, worker, QA, dashboard, MCP, install/start/bootstrap, or dogfood work; prefer the MCP onboard_route_guide tool before choosing a role or successor contract.
Public demo for Backlog Duplicate: a new requirement overlaps existing backlog work, and the observer asks whether to merge, supersede, or keep it separate. Use when a user asks to run, preview, or collect duplicate backlog evidence with the current Claude Code or Codex observer session.
Public HN challenge entrypoint for Aming Claw. Use when a user asks to run, preview, present, or collect evidence for the multi-agent challenge: one observer coordinates multiple commit-bound workers, records a failed or interrupted worker, replays it from the same contract lineage, reconciles the target graph, and writes an audit self-review.
HN demo case for the fear that code changes leave docs, tests, and config stale after implementation. Guides evidence collection for Asset Inbox, binding state, Baseline and Possible drift, Review Queue, impact scope, and review boundaries.
HN demo case for the fear that AI does not understand project structure and duplicates work. Guides evidence collection for graph-first discovery, backlog contract, target file fence, and acceptance criteria before implementation.
HN demo case for the fear that AI and subagents are black boxes during implementation. Guides evidence collection for timeline DAG, observer and subagent lanes, dispatch gate, evidence inspector, isolated worktrees, and append-only contract evidence.
| name | aming-claw-drift-demo |
| description | Public demo for Docs Drift: a feature changes but documentation is stale, and the drift state is surfaced before a second doc fix. Use when a user asks to run, preview, or collect evidence for docs drift with the current Claude Code or Codex observer session. |
Before any response that uses this skill, in this exact order:
ListMcpResourcesTool() ReadMcpResourceTool(uri="aming-claw://current-context") ReadMcpResourceTool(uri="aming-claw://skill") ReadMcpResourceTool(uri="aming-claw://graph-first")
current-context anchors project_id, governance URLs, and 3 guardrails. skill is the operating contract (Start Sequence, Observer Operating Modes). graph-first has copy-pasteable graph_query payload examples.
Common failures when these are skipped:
Show that a feature can change while docs are stale, and that the stale-doc state remains visible until a doc-fix step addresses it.
The preferred observer is the user's current Claude Code or Codex session. This session should run the two-step story: first change the feature while leaving docs stale, then guide the doc fix and compare the dashboard evidence.
Scripted runners are setup and CI fallback only. They may create an isolated fixture or run a deterministic smoke, but do not claim scripted output as proof that a live AI observer noticed documentation drift.
Use fixture setup only when there is no safe project ready.
Expected setup helper:
node frontend/dashboard/scripts/e2e-drift-demo-fixture.mjs --no-browser
Fixture setup should return a project id and prepare a small feature plus docs. It should not be described as the live observer's drift detection or product judgment.
Docs Drift evidence
- Project: <project_id>
- First feature change: <backlog id/link or timeline event>
- Stale docs signal: <asset/review/operations/graph evidence>
- Doc fix: <backlog id/link or changed doc>
- Dashboard: <graph link>, <assets link>, <review link>, <operations link>
- Limitations: <proposal not accepted/stale graph/scripted setup/etc>