fusion-dependency-review

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

• Third-party content exposure detected (high risk: 0.90). The skill's Research and Target-PR advisors explicitly require fetching and reading existing PR comments and review threads via "GitHub MCP" and upstream changelogs/issue trackers (e.g., agents/research-advisor.md and references/instructions.md), which are untrusted, user-generated third-party sources that the agent ingests and uses to drive lens assessments and final recommendations.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly requires fetching existing PR comments and review threads via GitHub MCP (e.g., https://api.github.com) at runtime to build the research packet that directly feeds advisor prompts, so this external API is a required runtime dependency that controls the agent's instructions.