plugin-research

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs the agent to fetch and parse open/public third-party content—e.g., GitHub repository pages (releases, tags, CHANGELOG.md, issues) and community resources like Reddit/discussions—via WebFetch and web searches, which are untrusted/user-generated sources that the agent is expected to read and interpret as part of its workflow.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs runtime WebFetch of GitHub pages (e.g., https://github.com/{owner}/{repo}/releases and https://github.com/{owner}/{repo}/blob/main/CHANGELOG.md) to extract release/CHANGELOG content that is injected into the agent's context to determine versions and breaking changes, so these external URLs are used at runtime to control the agent's prompts.