Scan your codebase for distinctive patterns — get structured scoring and evidence for patent consultation. NOT legal advice.
Security Analysis
high confidenceThe skill's inputs, instructions, and lack of installs/credentials are consistent with a local codebase analysis tool — nothing requested is disproportionate to its stated purpose.
Name/description (codebase patent-pattern scanning) align with instructions: the SKILL.md describes local repository discovery, prioritization, pattern analysis, and claim-angle drafting. It requires no binaries, env vars, or installs, which is proportionate for a pure-analysis skill.
The runtime instructions explicitly operate on a provided path or current directory and enumerate file types/dirs to include/exclude (expected). The SKILL.md also asserts "This skill operates locally. It does not transmit code or analysis results to any external service." That is a behavioral promise in the instructions (not an enforceable technical constraint) — the instructions themselves do not direct any external network calls or access to unrelated system files.
No install spec and no code files are present. Being instruction-only is low-risk and consistent with a static analysis helper.
The skill declares no required environment variables, credentials, or config paths. That matches the stated local analysis purpose and is appropriately minimal.
always is false and the skill does not request persistent presence or elevated privileges. disable-model-invocation is default (allows autonomous invocation), which is normal; there are no other privilege escalations requested.
Guidance
This skill appears internally coherent and is designed to run locally against a repository. Before installing or using it: 1) Confirm the agent runtime environment will not automatically transmit files or outputs to external services — the SKILL.md promises local-only operation but cannot enforce platform-level network behavior. 2) Avoid running the skill on sensitive/proprietary code unless you trust the agent runtime and logs; review any outputs it generates before sharing. 3) Remember the tool is explicitly non-legal and intended to surface technical patterns; consult a qualified patent attorney before taking IP-related actions. If you need stronger guarantees about data staying on your machine, run analysis in an isolated/offline environment or inspect agent/network policies first.
Latest Release
v1.4.0
- Removed the "disable-model-invocation: true" property from SKILL.md to allow model invocation. - No logic or behavioral changes; SKILL.md updated only for metadata adjustment.
More by @leegitw
Published by @leegitw on ClawHub
