Policy Delta Watcher
v1.0.0比较新旧制度或政策差异，指出业务影响、流程影响和需更新的操作手册。；use for policy, diff, governance workflows；do not use for 给法律定性结论, 忽略边缘影响人群.
⭐ 0· 104·0 current·0 all-time
byvx：17605205782@52yuanchangxing
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description match the included resources and script. The bundle includes templates, a spec.json describing sections, examples, and a python script that produces structured reports and directory/csv/pattern audits — all coherent with a policy-diff/briefing tool.
Instruction Scope
SKILL.md instructs the agent to run the included local script or to produce output from the bundled template; the script reads text files, directories, and CSVs and can scan a provided path for patterns and headings. This is expected for an audit tool, but it means the script will read any files/directories the user points it at (including .py, .sh, .md, .json). Users should avoid pointing the script at broad system roots or sensitive directories unless they intend to audit them.
Install Mechanism
No install spec; only a requirement of python3 and no third-party packages. The script is bundled in the repo and no external downloads or package installs are performed.
Credentials
The skill requests no environment variables, no credentials, and declares only python3 as a binary requirement. There are no unrelated secrets or cloud credentials requested.
Persistence & Privilege
always is false and the skill does not request permanent presence or modify other skills/configs. It can be invoked by the agent (normal), and its local script can write an output file per CLI args; this is proportional for a reporting tool.
Assessment
This skill appears to do what it says: generate structured briefs and local audits using the bundled Python script and templates. Before running: (1) review scripts/run.py (it's included and readable) to confirm behavior; (2) run in a safe/sandboxed directory or use --dry-run to avoid writing output files; (3) do not point the tool at system root or directories containing secrets unless you intend to audit them; and (4) because the source is 'unknown' and homepage is an example domain, prefer to run it locally in an isolated environment (or inspect the files) before granting it access to any sensitive inputs. There are no network/exfiltration calls or credential requests in the bundle.Like a lobster shell, security has layers — review code before you run it.
latestvk977dxpnmmvv78r7e5h52n9nqh836ktn
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
📜 Clawdis
OSmacOS · Linux · Windows
Binspython3