mem-clean
Pass
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes external maintenance scripts (
mem-sync-git.sh) and shell utilities likediffto manage repository synchronization, history compaction, and branch resolution. - [PROMPT_INJECTION]: The skill demonstrates an indirect prompt injection surface because it reads and parses the content of memory logs and conflict copies to identify lifecycle markers such as
[Candidate]or[Promoted]. - Ingestion points: Accesses and parses files located in
~/.agents/memories/and project-specific.memories/directories. - Boundary markers: Instructions do not specify the use of clear delimiters or guardrails to prevent log content from being interpreted as agent instructions during parsing.
- Capability inventory: Includes the ability to delete local files, merge data into primary files, and execute git force-pushes to remote repositories.
- Sanitization: No explicit sanitization or validation of the log content is described before the content influences the cleanup logic.
Audit Metadata