intake
Pass
Audited by Gen Agent Trust Hub on Oct 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process 'shabda' (external untrusted data) from various sources, creating a significant attack surface for indirect prompt injection where instructions hidden in external data could influence agent behavior.
- Ingestion points: The instructions specify processing external content from GitHub issues, RFCs, README files, source code fragments, and conversation transcripts.
- Boundary markers: There are no explicit instructions for using delimiters, XML tags, or other boundary markers to isolate the untrusted external text from the agent's internal instruction context.
- Capability inventory: The skill utilizes the iskron tool suite (iskron_semantic_search, iskron_case, iskron_look, iskron_orient) which allows for reading from and writing to a persistent knowledge graph and case management system.
- Sanitization: No technical sanitization, escaping, or filtering of the external input content is mandated before the agent processes or stores it.
Audit Metadata