gitnexus-exploring

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

• Third-party content exposure detected (high risk: 0.85). Steps 1–5 read from gitnexus://repos / gitnexus://repo/{name}/context / .../process/{name} which are codebase contents (source files, execution traces, symbol context) authored by parties other than the operating user, and those readable texts are ingested into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 1.00). This skill issues runtime READ requests to gitnexus:// URLs (e.g., gitnexus://repo/{name}/context and gitnexus://repo/{name}/process/{name}) to fetch repository context and execution traces that are injected into the agent's context and therefore directly control prompts/responses.