技能详情（站内镜像，无评论）
作者：Andy @andyylin
许可证：MIT-0
MIT-0 ·免费使用、修改和重新分发。无需归因。
版本：v1.0.0
统计：⭐ 0 · 203 · 0 current installs · 0 all-time installs
⭐ 0
安装量（当前） 0
🛡 VirusTotal ：良性 · OpenClaw ：良性
Package：andyylin/durable-files-weekly-review-public
安全扫描（ClawHub）
- VirusTotal ：良性
- OpenClaw ：良性
OpenClaw 评估
The skill's code and instructions match its stated purpose: it locally scans specific durable files, generates a markdown report, and requires explicit approval before any deletions; there is no network I/O or secret access requested.
目的
Name/description, SKILL.md, and the included Python script are consistent: the script scans the listed durable files, estimates token counts, detects stale markers, and writes a report. The skill does not request unrelated credentials, binaries, or config paths.
说明范围
Runtime instructions are narrowly scoped (run the included Python scanner on the workspace root and review its report). Note: the CLI accepts a --root argument, so whoever runs the tool can point it at any filesystem path; ensure the root parameter is set to the intended workspace to avoid scanning unrelated directories.
安装机制
No install spec (instruction-only with a bundled script). Nothing is downloaded or executed from remote URLs. The script runs locally with standard Python and writes a markdown file to the specified output directory.
证书
No environment variables, credentials, or external endpoints are requested. The script only reads the specified filenames under the given root and writes a local report; this is proportionate to the stated purpose.
持久
The skill does not request always:true and does not modify other skills or global agent settings. It writes only its own report file and does not perform deletions without explicit user approval as stated in SKILL.md.
综合结论
This skill appears coherent and limited in scope: it reads the listed durable files in whatever root you point it at, generates a markdown report, and requires explicit approval before edits. Before running: (1) confirm the --root argument points to the intended workspace (avoid running against system or home directories), (2) review the generated report before applying any cleanup, and (3) consider running the script in a sandbox or test work…
安装（复制给龙虾 AI）
将下方整段复制到龙虾中文库对话中，由龙虾按 SKILL.md 完成安装。
请把本段交给龙虾中文库（龙虾 AI）执行：为本机安装 OpenClaw 技能「Durable Files Weekly Review (Public)」。简介：Run a weekly token-optimization audit for durable instruction files in any Open…。
请 fetch 以下地址读取 SKILL.md 并按文档完成安装：https://raw.githubusercontent.com/openclaw/skills/refs/heads/main/skills/andyylin/durable-files-weekly-review-public/SKILL.md
（来源：yingzhi8.cn 技能库）SKILL.md
---
name: durable-files-weekly-review-public
description: Run a weekly token-optimization audit for durable instruction files in any OpenClaw workspace, generate a markdown report, and propose approval-gated cleanup actions. Use when users want to keep AGENTS/USER/TOOLS/MEMORY-style docs lean without silent deletions.
---
# Durable Files Weekly Review (Sanitized / ClawHub)
Use this skill to audit durable instruction files and prepare cleanup proposals.
## Default scope
Audit these files relative to a workspace root:
- `AGENTS.md`
- `SOUL.md`
- `USER.md`
- `TOOLS.md`
- `MEMORY.md`
- `IDENTITY.md`
- `PRIORITIES.md`
- `SKILLS.md`
- `projects.md`
## Run
```bash
python3 scripts/durable_files_review_generic.py --root .
```
Optional:
```bash
python3 scripts/durable_files_review_generic.py --root /path/to/workspace --out knowledge/reports/durable-files
```
## Workflow
1. Run scanner.
2. Open generated report.
3. Summarize top token-heavy files + stale markers.
4. Propose cleanup in batches.
5. **Require explicit user approval before any deletions.**
6. Apply approved edits and post concise changelog.
## Success output
- Dated markdown report with metrics/findings
- Clear approval queue for removals
- No silent content removal