Skillv0.1.0

ClawScan security

Semantic Paper Radar · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

ReviewMar 3, 2026, 12:00 PM

Verdict: Review
Confidence: medium
Model: gpt-5-mini
Summary: The skill's code and instructions are mostly coherent with a literature-discovery tool, but the provided script was truncated in the bundle preview and there are a few minor operational/privacy concerns (unencrypted arXiv fetch, external network calls, optional cross-skill invocation) that merit caution before installation.
Guidance: This skill appears to do what it says — it queries OpenAlex, PubMed, and arXiv and ranks papers — but take these precautions before installing or enabling it: - Review the full, untruncated script on disk (the preview was truncated). The analysis above used the visible portion; any hidden/truncated code could change this assessment. - Be aware queries are sent to external public APIs (api.openalex.org, eutils.ncbi.nlm.nih.gov, export.arxiv.org). Do not send private or patient-identifiable data through this skill. - Note: the arXiv query uses HTTP (export.arxiv.org) in the visible code (unencrypted); if confidentiality is important, update to HTTPS or confirm arXiv's recommended endpoint. - The tool can export HTML to a path you control; confirm file paths before exporting to avoid accidental overwrites. - The SKILL.md suggests calling an existing Scholar-capable tool if present — ensure any cross-skill calls don't expose tokens/credentials belonging to other skills. - If you require higher assurance, run the script in a sandboxed environment or review the entire file contents locally to ensure there is no unexpected network endpoint or data-exfiltration behavior.

Review Dimensions

Purpose & Capability: okThe SKILL.md describes semantic literature discovery across arXiv/OpenAlex/PubMed and the included script implements those exact network queries and ranking logic. Required resources (none) align with the stated purpose.
Instruction Scope: noteRuntime instructions tell the agent to run the included Python script and optionally call a Scholar-capable tool if present. The workflow intentionally performs network retrievals from public APIs and can export HTML locally. This is in-scope for a literature discovery skill, but it means user queries and terms will be sent to external services; SKILL.md also forces Chinese output by default—an operational/policy choice to be aware of.
Install Mechanism: okNo install spec is present (instruction-only with an included script). Nothing is downloaded or installed automatically by the skill manifest.
Credentials: okThe skill requests no environment variables or credentials. All external calls use public APIs (OpenAlex, PubMed, arXiv) and no secret material is required by the code shown.
Persistence & Privilege: okalways is false and model invocation is normal. The skill does not request persistent platform privileges in the manifest. It may write an exported HTML file if the user requests --export-html / --html-out, which is reasonable for this functionality.