XDOG

Tools for building AI agents and managing LLM deployments.

GitHub →

xdog / Packages / skills / Overview

Goal and plan instructions, not another Agent harness.

The common skills hub ships goal and plan, plus isolated Goal domain support and an optional readonly display extension. Flow and Eyes keep their own engines and packaged skills. There is no separate xdog-goal distribution or xdog-skills management CLI.

Highlights

  • Coding and Claw include xdog-skills by default.
  • Skill metadata is discoverable; full bodies are not activated by default.
  • Goal model tools and native /goal commands are independent of body activation.
  • The current main Agent plans, works/delegates, reports progress and judges completion.
  • Goal owns records/task files and signal-driven reminders, not a worker/reviewer runner.
  • One current Goal per stable owner; completed remains visible until explicit clear or a new Goal.
pip install xdog-skills

Use the commands inside a capability-enabled Coding/Claw session, not as shell commands. A generic SDK host must open the appropriate installed capability with real owner/ACP/lifecycle ports; installing a skill does not invent those ports.

See Reference and the packaged instructions.