gitnexus-exploring

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md’s required workflow reads repository-derived resources via gitnexus://repos / gitnexus://repo/{name}/context / gitnexus://repo/{name}/process/{name}, which at runtime would ingest outsider-authored codebase text/metadata (e.g., staleness/context traces and execution-flow descriptions) into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill issues runtime READ calls like gitnexus://repo/{name}/context (and related gitnexus://... endpoints) which are fetched at runtime and directly injected into the agent's context to control its responses, so these external URLs represent required runtime content that can control prompts.