The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on the dce CLI tool and a Python script (scripts/collect_rca.py) to gather diagnostic data.
Evidence: The script uses subprocess.run with argument lists (e.g., subprocess.run(list(command), ...)) to execute various dce subcommands such as insight tracing query-spans and container-management cluster list-clusters. This method is secure against shell injection.
[PROMPT_INJECTION]: The skill ingests data from external observability sources (Jaeger traces, pod logs, and system alerts), which represents an attack surface for indirect prompt injection.
Ingestion points: The script scripts/collect_rca.py retrieves logs and trace data using the dce command-line utility.
Boundary markers: No explicit delimiters or instructions are used to isolate ingested data from the agent's core instructions.
Capability inventory: The skill has the capability to read system-wide telemetry and logs but lacks file-writing or general outbound network permissions.
Sanitization: The script does not perform sanitization on the text content of logs or trace tags, relying on the agent's internal safety filters.

openclaw-request-rca