openclaw-diagnose-gateway
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from logs and tool outputs that could potentially contain malicious instructions.
- Ingestion points: System log files (e.g.,
/tmp/openclaw/openclaw-<date>.log), application channel logs (openclaw channels logs), and diagnostic reports from theopenclaw doctorcommand. - Boundary markers: The instructions do not define explicit delimiters or specific 'ignore' directives to isolate log content from the agent's instructional context.
- Capability inventory: The skill enables the agent to execute shell commands including the
openclawCLI, system utilities likelsof, and repository-local scripts (./scripts/clawlog.sh). - Sanitization: Log data and tool outputs are processed directly without sanitization or escaping mechanisms.
- [COMMAND_EXECUTION]: The skill utilizes several system and application-specific commands for diagnostic purposes.
- Evidence: Uses
lsof -nP -iTCP:<port> -sTCP:LISTENto inspect network listeners and various subcommands of theopenclawCLI (doctor,gateway status,config validate,channels status,models status) to verify system state.
Audit Metadata