The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is instructed to read the primary openclaw.json configuration file and scan the memory/ directory for API keys using grep -r "sk-", which provides access to sensitive credentials and environment data.
[COMMAND_EXECUTION]: The agent is provided with shell commands (using sqlite3, jq, and sed) to perform "auto-fixes" on the system, including modifying core configuration files and manipulating the internal database.
[COMMAND_EXECUTION]: The diagnostic process includes commands to identify and terminate processes using specific ports (lsof -ti:3000 | xargs kill -9), granting the agent process management capabilities.
[EXTERNAL_DOWNLOADS]: The skill's documentation and installation procedures suggest fetching remote scripts and resources via curl, wget, and git clone from external sources.
[COMMAND_EXECUTION]: The provided auto-diagnostic.sh script utilizes eval to execute dynamically constructed command strings during the health check phase.

agent-doctor