Skillv1.0.2

ClawScan security

AI-first pastebin · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

BenignApr 30, 2026, 3:34 AM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill is internally consistent: it bundles a small Python CLI that talks to the 呵.tw (xn--dtr.tw) pastebin service and its instructions match the code and claimed purpose; it does transmit user-provided content to an external host (expected for a pastebin).
Guidance: This skill appears to do what it says: a small Python CLI that sends non-secret text to the 呵.tw (xn--dtr.tw) pastebin service. Before installing, consider: (1) never store secrets — the skill will transmit content to an external server; (2) verify you trust the service operator and review any privacy/retention policy at https://呵.tw/llms.txt or the service's website; (3) if you fear accidental exfiltration, avoid enabling autonomous invocation or require explicit user approval before the agent runs the skill; (4) you can inspect the included script (scripts/hotw.py) — it’s short and readable and performs only HTTPS GET/POST calls — or run it in a restricted environment if you want extra assurance.

Purpose & Capability: okName/description (AI-first pastebin) match the code and SKILL.md: the included hotw.py implements shortening, paste creation, resolution, metadata, chain, find and QR helpers against https://xn--dtr.tw (displayed as https://呵.tw). No unrelated binaries, env vars, or config paths are requested.
Instruction Scope: noteSKILL.md instructs the agent to run the bundled Python CLI and explicitly warns not to upload secrets. The CLI reads stdin or arguments and sends content to the remote API endpoints (/api/create, /api/paste/create, /api/resolve, etc.). This is expected for a pastebin but it does mean user content is transmitted off-host — the guidance to avoid secrets is appropriate and should be followed.
Install Mechanism: okNo install spec; this is instruction-only with an included Python script. Nothing is downloaded or written to disk beyond the skill bundle; runtime requires only python3 which is normal and proportionate.
Credentials: okThe skill requests no environment variables, no credentials, and no config paths. There are no hidden credential requirements in SKILL.md or the script. The only external interaction is HTTPS requests to the pastebin service.
Persistence & Privilege: okalways is false, user-invocable is true, and the skill does not modify other skills or system-wide settings. It does allow agent invocation (normal default); there is no evidence of privilege escalation or persistent background processes.