team-health
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a persistent context mechanism that reads from and writes to local files, creating a surface for indirect prompt injection where malicious input could be saved and later re-processed as instructions.
- Ingestion points: The skill reads from
.agents/em-context.mdand.agents/reports/[name].md(inSKILL.md) to initialize its operational context. - Boundary markers: No delimiters or protective instructions are used to separate the stored file content from the agent's core instructions.
- Capability inventory: The skill relies on standard file system read and write operations via default agent tools to manage its context files.
- Sanitization: There is no programmatic sanitization or validation of the data before it is saved to these persistent files or when it is subsequently read into the active prompt context, though it provides qualitative guidance to the agent to save only stable facts.
Audit Metadata