continuous-learning
Pass
Audited by Gen Agent Trust Hub on Mar 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides a shell script
evaluate-session.shmeant to be registered as a 'Stop' hook in the agent's environment. This script uses standard utilities (grep,sed,jq) to calculate session length and message counts from transcripts. - [DATA_EXFILTRATION]: While the script reads session transcripts, it does so locally to evaluate whether a session meets the criteria for pattern extraction. The data is processed within the user's local filesystem (
~/.claude/skills/learned/) and is not transmitted to external servers. - [PERSISTENCE]: The skill instructs the user to modify
~/.claude/settings.jsonto add a persistent hook. This hook triggers the evaluation script automatically at the conclusion of every session, which is the intended core functionality of the skill.
Audit Metadata