← Lab Experiment

Documentation

Docs for humans. Skills for agents.

These pages are the human version — context for the questions you'll ask your agent. Your agent can fetch the live version directly at ouroboros.bot/skills/manifest.json

Words used here
Psyche
Five Markdown files that define who the agent is, loaded every turn. More →
Bundle
One folder holding an agent's identity, memory, and tools, so you can back it up and move it. More →
Reorientation overhead
What it costs, usually the human, to get an agent back on track after it loses the thread. More →

Docs for humans, skills for agents. The website explains the what. Your agent fetches the how from the code, the local install, or the bundle it's working inside.

Have a question?

Ask your agent 
Inspect the latest local Ouroboros install or repo on this machine and answer this using the current source of truth: [your question here]

Each docs page also has a more specific prompt. This one works for anything.

  • Run one command. Talk to the serpent.

    npx ouro.bot starts an interactive guide that walks you through provider setup, naming, and personality.

  • Your agent has a portable self.

    What lives in a bundle — psyche, diary, journal, habits, friends — and how to ask your agent to inspect them.

  • Keep the thread.

    The diary is the shelf. The journal is the desk. Habits are the rhythm. How they work together.

  • Understand the body.

    Modules named like organs. Habits, inner dialog, MCP bridge, and a runtime the model can read and extend.

  • A place to put the work.

    Task cards, work directories, a compact board in every prompt, and self-repair tooling when things drift.

  • Hand work off cleanly.

    Prompts for quick questions. Skill handoffs for tasks that need the latest local truth.

  • Start from first principles.

    What an agent harness is, why it matters, and how it differs from a framework.