The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple shell commands to manage the OpenClaw environment, including openclaw agents add, openclaw gateway restart, and a bundled Python script scripts/upsert_openclaw_agent.py to update the local configuration file.\n- [CREDENTIALS_UNSAFE]: The skill collects and processes sensitive information such as app_secret. These credentials are passed as command-line arguments to the upsert_openclaw_agent.py script, which can expose them to other users or monitoring tools on the same host through process list inspection.\n- [PROMPT_INJECTION]: The skill is subject to indirect prompt injection through the monitoring of external logs to extract group identifiers.\n
Ingestion points: Standard output from the openclaw logs --follow command, which reflects messages sent from external messaging platforms (Feishu).\n
Boundary markers: The skill does not implement specific boundary markers or 'ignore' instructions when reading the log stream, relying on the agent to manually identify patterns.\n
Capability inventory: The skill has capabilities to execute shell commands, read/write configuration files, and restart system services.\n
Sanitization: The Python script upsert_openclaw_agent.py includes regex validation for the agent_id parameter to ensure it matches expected alphanumeric patterns.

openclaw-create-agent