research-blockchain

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

• Third-party content exposure detected (high risk: 0.90). The skill's SKILL.md requires the agent to run Zai MCP web tools (mcp__web-search-prime__webSearchPrime, mcp__web-reader__webReader pointing at https://docs.${PROTOCOL}.io, and mcp__web-search__fetchGithubReadme for GitHub), and cites public sites (DefiLlama, Etherscan, Dune), so it explicitly fetches and ingests untrusted public/web and user-generated content which the agent is expected to read and use to drive research decisions.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill's runtime tool definitions fetch and inject external documentation into the agent context—specifically URLs like "https://docs.${PROTOCOL}.io" and "https://github.com/${ORG}/${REPO}" via mcp__web-reader__webReader and mcp__web-search__fetchGithubReadme—so external content can directly control prompts and is relied on for the agent's outputs.