caira

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs the agent to inspect and fetch live content from the public GitHub CAIRA repository (e.g., via the GitHub contents API GET https://api.github.com/repos/pablozaiden/CAIRA/contents/... and repo file download URLs) and to read/interpret those files to drive architecture, code generation, and deployment decisions, exposing it to untrusted third-party content that can influence actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs the agent to fetch and ingest CAIRA repository files at runtime via the GitHub contents API (e.g., https://api.github.com/repos/pablozaiden/CAIRA/contents/deployment-strategies?ref=main), using those fetched files as the source-of-truth that guides prompts, recommendations, and generated code, which makes this an external runtime dependency that directly controls agent behavior.