sourcebot
Pass
Audited by Gen Agent Trust Hub on Jun 27, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses coercive language to override the agent's natural decision-making process. Specifically, it mandates that the
sourcebottool must be used exactly "5-7 times" and threatens "TASK INVALIDATION" if this arbitrary quota is not met. - [PROMPT_INJECTION]: The instructions provide an attack surface for indirect prompt injection by processing untrusted data from external repositories.
- Ingestion points: External code and documentation are fetched via the
sourcebotsearch andget_file_sourcetools. - Boundary markers: The instructions do not define delimiters or XML tags to isolate external content from the system prompt.
- Capability inventory: The agent has access to external search tools (Sourcebot, Perplexity, Context7) and local codebase search tools.
- Sanitization: There are no instructions for validating, escaping, or filtering the content retrieved from external repositories before processing or displaying it.
- [PROMPT_INJECTION]: The skill instructs the agent to use specific third-party services ("Perplexity and Context7") after completing its primary task, which may steer the agent away from the user's intended workflow or the platform's preferred tools.
Audit Metadata