The Agent Skills Directory

[SAFE]: The skill's primary purpose is to provide educational and auditing guidelines for secure Ruby development. All instructions and scripts are consistent with this purpose.
[PROMPT_INJECTION]: The skill is designed to analyze external, untrusted codebases in 'AUDIT mode', creating a potential surface for indirect prompt injection. Malicious instructions within audited project files could potentially be interpreted by the agent. This risk is inherent to the functionality of a code auditing tool.
Ingestion points: The skill reads repository files such as .rb, Gemfile, and .ruby-version during auditing.
Boundary markers: There are no explicit instructions to delimit or ignore instructions found within the analyzed project files.
Capability inventory: The skill uses standard search and analysis tools like grep, cat, and bundle on the ingested data.
Sanitization: Project file content is processed directly for pattern matching without specific sanitization steps.
[EXTERNAL_DOWNLOADS]: The skill includes instructions to use bundler-audit to update a local vulnerability database. This operation fetches data from a well-known community resource and is a standard security practice.

sota-ruby