The Agent Skills Directory

[SAFE]: The skill consists entirely of markdown documentation providing technical specifications for the OpenClaw platform. No executable scripts, binaries, or active logic are present.
[NO_CODE]: As a documentation-only skill, it lacks any code execution surface. All instructions are descriptive and intended for use as reference context by an AI agent.
[SAFE]: Documentation identifies architectural system paths such as ~/.openclaw/credentials/ and ~/.openclaw/openclaw.json. No hardcoded secrets, actual credentials, or sensitive configuration data are exposed within the skill files.
[COMMAND_EXECUTION]: The skill provides an exhaustive list of CLI commands and provisioning workflows (e.g., using curl and jq to merge configuration) intended for system administrators. These are presented for reference and do not contain malicious payloads or execute code within the skill context.
[EXTERNAL_DOWNLOADS]: Technical specifications describe the platform's plugin installation mechanism using npm, ClawHub, and archives. This information is purely descriptive of the platform's capabilities.
[PROMPT_INJECTION]: No instruction overrides or attempts to bypass safety filters were detected in the documentation text or metadata.

openclaw-ref