The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/openclaw_config.py file contains logic to restart the OpenClaw gateway service using subprocess.run. It attempts to execute hardcoded commands such as systemctl restart openclaw-gateway. This is an intended administrative feature of the skill, and since the commands are defined as static lists, the risk of command injection is mitigated.
[DATA_EXPOSURE]: The skill is designed to manage openclaw.json files, which may contain sensitive information like apiKey fields as described in the architecture documentation. The skill performs local file operations (reading, writing, and creating timestamped backups in config-backups/) to facilitate configuration management. No network operations were found that would indicate data exfiltration.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from local JSON configuration files and model definition files. This creates a surface for indirect prompt injection, as an agent might follow instructions embedded in a malicious config file. However, the skill implements specific JSON schema checks and lacks general-purpose network tools, which reduces the exploitability of this vector.

openclaw