query

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

• Third-party content exposure detected (high risk: 0.80). This skill's SKILL.md directs the agent to execute read-only queries against the public Convex network using the mcp__convex-testnet__query tool, so it ingests untrusted, user-generated on-chain content that the agent must read and act on, which could materially influence its outputs.