The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates administrative tasks through the openclaw CLI, enabling the management of cluster gateways and agent lifecycle operations.\n- [PROMPT_INJECTION]: The skill contains a vulnerability surface for indirect prompt injection via the processing of external configuration files.\n
Ingestion points: The openclaw config patch --file command reads data from external JSON files (SKILL.md).\n
Boundary markers: No specific delimiters or instructions are provided to the agent to treat file-based content as untrusted or to ignore instructions embedded within the patches.\n
Capability inventory: The skill can execute local system commands and perform network operations.\n
Sanitization: No logic is documented for validating or sanitizing the content of the configuration files before they are processed by the CLI tool.\n- [EXTERNAL_DOWNLOADS]: The openclaw gateway update command allows for fetching and installing specific versions of the gateway software from vendor repositories.\n- [DATA_EXFILTRATION]: The skill uses curl to make network requests to api.openclaw.com for gateway management and status checks.\n- [CREDENTIALS_UNSAFE]: The documentation describes the use of the OPENCLAW_API_KEY environment variable for authentication and provides standard placeholder examples for its configuration.

openclaw-admin