graphify

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

• Third-party content exposure detected (medium risk: 0.30). In Step 3 Part B, when semantic extraction is run via the subagent tool (OpenCode @agent Chunk ...), the agent is fed the contents of untrusted, user-supplied corpus files (docs/papers/images) derived from the input path/URLs, enabling outsider-authored text to influence the LLM.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill accepts and clones/fetches arbitrary GitHub repositories at runtime (e.g. https://github.com//), and fetched files are then ingested and read by semantic subagents (via the Agent tool) so external content can be injected into agent prompts and influence behavior.