self-improving-agent
Pass
Audited by Gen Agent Trust Hub on Mar 11, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses standard local CLI tools such as grep, sed, ls, wc, and find to identify patterns in project memory and verify the existence of referenced files. These operations are conducted within the user's project environment and home directory configuration paths.
- [SAFE]: The skill implements a PostToolUse hook to monitor command outputs for errors. This functionality is contained within a local bash script and does not perform network operations or access data outside the project context.
- [SAFE]: The skill facilitates the modification of agent instructions (CLAUDE.md and .claude/rules/) through a guided promotion workflow. This process is designed to improve agent performance by graduating learned patterns into enforced project rules, involving user confirmation at key steps.
Audit Metadata