opik
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill documents architectural patterns for building agents that ingest and process external content from untrusted sources such as web pages or retrieved documents.
- Ingestion points: Specifically addressed in
references/agent-patterns.mdwhere input queries and retrieved context are passed to the agent entry points and LLM spans. - Boundary markers: The documentation recommends using explicit boundary markers and input validation to separate instructions from data.
- Capability inventory: The skill facilitates the integration of LLMs with external tools (API calls, search, database retrieval) as shown in
references/integrations.mdandreferences/agent-patterns.md. - Sanitization: The skill provides specific remediation code in
references/agent-patterns.md, demonstrating how to use regex to strip instruction-like patterns (e.g., 'ignore previous', 'system:') from retrieved content to prevent injection.
Audit Metadata