菜单
How to record completed work in logs/agent-log.md so humans can inspect what the agent read, wrote, and concluded.
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。
Required procedure for answering questions about the client - grounding, citation format, conflict handling, and how to say "not in the workspace".
How to create deliverable artifacts (meeting prep briefs, draft emails, analyses) in artifacts/ with required structure and source manifests.
How to extract durable memories from sources and write them as reviewable candidate memory files with provenance, confidence, and sensitivity.
How to orient inside a client workspace - the directory layout, the manifest, provenance records, and which sources outrank which.
| name | audit-logging |
| description | How to record completed work in logs/agent-log.md so humans can inspect what the agent read, wrote, and concluded. |
Every completed task gets one entry appended to logs/agent-log.md. Tool calls are logged automatically to logs/tool-calls.jsonl; the agent log is the human-readable layer on top: what you did and why.
Append exactly one entry after finishing a task (successful or not), as your final action before giving your final answer. Use writeFile with append: true.
## <ISO timestamp> — <one-line task summary>
- task: <the request, restated in one line>
- read: <comma-separated list of files you relied on>
- wrote: <comma-separated list of files you created/updated, or "none">
- conclusion: <1-3 lines: outcome, key findings, or why the task could not be completed>
- caveats: <conflicts found, missing data, low-confidence items, or "none">
Keep the entry under 15 lines. The log is append-only: never rewrite or delete previous entries.