context7

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

• Third-party content exposure detected (high risk: 0.85). Context7’s runtime workflow performs GET requests to https://context7.com/api/v1/search and https://context7.com/api/v1/{libraryId}?topic=..., ingesting the returned markdown/JSON docs from an external service (outsider-authored documentation) into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill performs runtime GET requests to https://context7.com/api/v1 (e.g., /search and /{libraryId}?topic=...) and injects the returned documentation into the agent's responses, so externally-fetched content directly controls the model's prompts/outputs and is a required runtime dependency.