Back to skill
Skillv1.0.0
ClawScan security
zhipu web fetch · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignMar 14, 2026, 1:16 AM
- Verdict
- benign
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- The skill's requests, instructions, and files are consistent with a simple web-page fetcher that posts a URL to Zhipu's Reader API; nothing in the package appears disproportionate or covert.
- Guidance
- This skill legitimately forwards the URL and fetch options to Zhipu's Reader API at open.bigmodel.cn, so avoid using it with private/internal URLs or pages containing sensitive data you don't want sent to a third party. Keep your ZHIPU_API_KEY secret and scoped appropriately. Confirm the endpoint (open.bigmodel.cn) is acceptable for your data governance, and consider network egress rules or monitoring if you need to prevent uploads of internal URLs. Otherwise the package is straightforward and matches its stated purpose.
Review Dimensions
- Purpose & Capability
- okName/description, required binary (curl), required env var (ZHIPU_API_KEY), and the included shell script all align with a web-page fetching/reader integration for Zhipu's API.
- Instruction Scope
- okSKILL.md and the script only send the provided URL and explicit fetch options to the documented API endpoint; they do not read other files, environment variables, or system state beyond ZHIPU_API_KEY.
- Install Mechanism
- okNo install spec; the skill is instruction-only with a simple shell script. Nothing downloads/extracts or installs third-party code.
- Credentials
- okOnly ZHIPU_API_KEY is required and it is used directly in the Authorization header as expected. No unrelated credentials or high-privilege paths are requested.
- Persistence & Privilege
- okalways is false and the skill does not request system-wide changes. The default autonomous-invocation flag is present (normal) but not combined with other concerning privileges.