vs-rfc-research

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

• Third-party content exposure detected (high risk: 0.90). This skill explicitly instructs the agent to use octocode MCP to search and read public GitHub content (see SKILL.md Phase 3 and README which require githubSearchCode/githubGetFileContent/githubSearchRepositories), so it ingests untrusted, user-generated third-party content that directly informs decisions and RFC outputs.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill requires the octocode MCP tool to perform githubGetFileContent/githubSearch* calls at runtime (see the referenced octocode repo https://github.com/nicepkg/octocode), which will fetch arbitrary GitHub file contents and inject them into the agent context to drive prompts and the RFC — a runtime remote-content dependency that can directly control the agent.