persistent-memory-hygiene
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes historical session logs stored in
memory/YYYY-MM-DD.mdandMEMORY.md. These files serve as ingestion points for data that could potentially contain untrusted content (such as summaries of malicious websites) recorded during previous tasks. - Ingestion points: Historical session logs and state files accessed in
SKILL.md. - Boundary markers: There are no instructions to the agent to treat the contents of memory files as data rather than instructions, nor are there delimiters used.
- Capability inventory: The skill is restricted to reading and writing within its specific memory and state directories.
- Sanitization: The instructions do not include steps to sanitize or filter the content being reviewed during the hygiene process.
Audit Metadata