ruminate
Pass
Audited by Gen Agent Trust Hub on May 16, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local shell scripts and Python scripts to process data. This includes
sh .agents/skills/meditate/scripts/snapshot.shandpython3 .agents/skills/ruminate/scripts/extract-conversations.py. These operations are performed on local files within the agent's expected working directory and are used for the primary function of the skill. - [EXTERNAL_DOWNLOADS]: No external downloads or network requests were detected in the skill instructions or the Python script.
- [PROMPT_INJECTION]: No evidence of prompt injection or system instruction bypass patterns was found.
- [DATA_EXFILTRATION]: The skill reads local conversation history from
~/.claude/projects/, but there are no network commands (like curl or wget) that would send this data to an external server. The analysis is performed by sub-agents or sequentially on the local machine.
Audit Metadata