session-cleaner
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands to manage system processes and run internal scripts.
- Evidence: Usage of
pkill -9 -f "openclaw-gateway"andopenclaw gateway startinSKILL.mdto control the application state. - Evidence: Invocation of the local cleanup utility
python3 ~/clawd/skills/session-cleaner/scripts/clean_sessions.py. - [INDIRECT_PROMPT_INJECTION]: The skill processes session metadata and transcript files which may contain untrusted data.
- Ingestion points: Reads session data from
~/.openclaw/agents/main/sessions/sessions.jsonand scans*.jsonltranscript files inscripts/clean_sessions.py. - Boundary markers: No specific delimiters or warnings are used to prevent the agent from interpreting content within these files as instructions if they are displayed during the cleaning process.
- Capability inventory: The skill possesses the ability to kill processes, modify file names, and rewrite session configuration files.
- Sanitization: No sanitization is performed on the data loaded from the JSON files before the agent interacts with it.
Audit Metadata