context-discovery
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and process data from potentially untrusted external sources, such as third-party documentation and git history, which may contain malicious instructions designed to influence agent behavior.
- Ingestion points: Workflow steps 2, 3, and 6 in
SKILL.mddescribe reading data from git history (sem), past agent sessions (ctx), and external library documentation (context7). - Boundary markers: The instructions do not provide explicit delimiters or warnings to ignore instructions embedded within the retrieved context.
- Capability inventory: The skill primarily performs read-only operations (search, summary, query) to build context, which limits the immediate impact of an injection.
- Sanitization: There is no evidence of sanitization or filtering applied to the content retrieved by these tools before it is presented to the agent.
Audit Metadata