semantic-search

Pass

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: LOWSAFE
Full Analysis
  • [DATA_EXFILTRATION] (LOW): The skill contains commands to interact with local services via curl http://localhost:8081/v1/schema. While it performs network operations, they are restricted to the local loopback address, posing no risk of data exfiltration to external actors.
  • [INDIRECT_PROMPT_INJECTION] (LOW): The skill facilitates the ingestion of external data (source code from a codebase) into the agent's context via the search:semantic command.
  • Ingestion points: Results from codecompass search:semantic which retrieve code snippets from the local filesystem.
  • Boundary markers: None explicitly defined in the execution guide for the agent's output.
  • Capability inventory: Limited to searching and displaying code context; no destructive file-write or remote execution capabilities are present in this skill.
  • Sanitization: None described for the retrieved code content.
  • Risk Assessment: The severity is LOW as the content is used for discovery and reasoning rather than driving high-privilege side effects.
Audit Metadata
Risk Level
LOW
Analyzed
Feb 16, 2026, 08:51 AM