memory-compact
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest, classify, and rewrite persistent memory entries. Since these entries are sourced from prior interactions which may contain untrusted or adversarial content, there is a risk of instructions inside the memory being obeyed during the compaction process. The skill lacks documentation regarding sanitization or the use of specific boundary markers to isolate memory content from instructions.
Audit Metadata