Back to skill
Skillv1.0.0
ClawScan security
Patent Assistant · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignApr 17, 2026, 5:19 PM
- Verdict
- Benign
- Confidence
- medium
- Model
- gpt-5-mini
- Summary
- Instruction-only patent-drafting skill whose requested footprint matches its stated purpose, but it can produce unsupported prior-art or experimental claims if outputs aren't independently verified.
- Guidance
- This is an instruction-only patent-drafting assistant that appears internally consistent with its stated purpose and requests no credentials or installs. Key cautions: (1) The skill asks for prior-art lists and experimental data but does not identify data sources—LLMs can invent citations or measurements. Always verify any D1–D5 references in official databases (CNIPA, EPO, USPTO, Google Patents, Derwent) before relying on them. (2) Do not treat generated experimental data, test procedures, or affidavit templates as legally or scientifically authoritative—have a qualified attorney or technical expert review and sign any documents for filing. (3) Avoid pasting highly confidential trade secrets unless you have appropriate confidentiality controls. If you need the skill to perform actual database searches, prefer supplying the search results yourself or cross-check outputs against primary sources.
Review Dimensions
- Purpose & Capability
- okThe name/description (patent drafting, review, reply to office actions) aligns with the SKILL.md workflow (diagnosis, rewrite, packaging). The skill requests no binaries, installs, credentials, or config paths that would be unrelated to patent work.
- Instruction Scope
- noteThe instructions are focused on patent tasks (novelty search, drafting, claim rewriting, reply templates). However, the skill asks for '5 closest prior-art documents (D1-D5)' and recommends providing experimental data and test methods without specifying data sources or verification steps. That leaves room for the agent to hallucinate cited prior art or fabricate supporting data—users should verify any cited references and experimental claims independently. Also, outputs that are legal statements or declarations (e.g., affidavits/宣誓书) should be reviewed by a qualified attorney before use.
- Install Mechanism
- okNo install spec and no code files; it's instruction-only so nothing will be downloaded or written to disk by an installer.
- Credentials
- okThe skill declares no environment variables, credentials, or config paths. It does not request access to unrelated services or secrets.
- Persistence & Privilege
- okalways:false (default). The skill does not request permanent/system-wide presence or elevated privileges. Autonomous invocation is allowed by platform default but is not combined here with broad access to credentials or system config.