Security audit
Security checks for vulnerabilities and agentic risk
This skill coherently connects an agent to the RegMapper API for regulatory research and project annotations, with its read and write capabilities disclosed and scoped by the user's RegMapper token permissions.
Install this only if you trust RegMapper with the projects your API token can access. Use a least-privilege token where possible, keep REGMAPPER_API_TOKEN out of source control, and review agent actions before creating or updating shared project comments because those changes persist in RegMapper.
The comment endpoints can return full note content, including project annotations that may contain sensitive internal compliance discussions or personal data. In an agent setting, insufficiently explicit privacy signaling increases the chance the model retrieves and surfaces confidential note text to the user when only summaries or metadata were needed.
The manifest says the skill is for browsing, searching, and annotating regulations, which strongly implies read-oriented access plus annotation creation. However, the OpenAPI spec also exposes PATCH-based comment updates in addition to creation, meaning the skill can modify existing remote project records, not just add annotations. That is a meaningful behavior expansion beyond the simpler description.
The POST and PATCH comment endpoints perform data-changing operations by creating or altering stored comments, notes, statuses, or categories. Although permissions are noted, the documentation does not explicitly warn that these actions change persisted project data and may affect collaboration records or audit-sensitive content.
The manifest focuses on regulations, compliance requirements, annotations, and comparing regulation versions. The schemas for comparison responses additionally provide orig_url and copyright_url fields, which surface external document and copyright links unrelated to the core annotation/search workflow. This is a supplementary capability not reflected in the stated purpose.
No suspicious patterns detected.
