The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill explicitly uses administrative privileges via the sudo command to execute deployment scripts such as apply-executor-slice.sh. This is used to install and modify system-wide LaunchDaemons on the host machine.
[EXTERNAL_DOWNLOADS]: The skill instructs the agent to perform a git clone of the repository https://github.com/RhysSullivan/executor. This external source is used to inspect upstream source code and verify configuration patterns.
[COMMAND_EXECUTION]: The agent is directed to execute numerous system-level commands, including launchctl for service management, lsof for network port monitoring, pgrep for process identification, and curl for performing health checks against local and Tailscale network endpoints.
[PERSISTENCE]: The core functionality of the skill involves creating and managing launchd service configurations (com.joelclaw.central.executor), which establishes persistence by ensuring the executor daemon starts automatically during the system boot process.
[CREDENTIALS_UNSAFE]: The skill identifies and interacts with a sensitive authentication storage path at /Users/Shared/joelclaw/data/executor/server-control/auth.json. Although the instructions contain explicit warnings against printing or exposing the contents of this file, the skill provides the agent with the exact location of the service's bearer tokens.

executor