proof

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs the agent to extract a shared "token" from URLs and include it verbatim in Authorization headers and curl commands, requiring the LLM to handle and emit secret values.

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 join and read Proof documents and comment threads from the public Proof API (e.g., GET https://www.proofeditor.ai/api/agent//state and /snapshot), ingesting user-generated content that the agent then uses to decide edits and actions, so untrusted third-party content can influence behavior.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly fetches live document state and snapshots at runtime from https://www.proofeditor.ai/api/agent//state (and related endpoints like /snapshot and /edit/v2), and that fetched content is read into the agent and used to drive its actions and outputs, so remote content directly controls prompts/instructions.