Repolode
Submit
ray-r-ren

ray-r-ren/agent-apprenticeship

A CLI that runs AI coding agents through iterative task loops and turns their completed work into reusable experience data for future agents.

View on GitHub
Stars
1,617
Forks
61
Open issues
0
Last commit
2 days ago

What is it?

Agent Apprenticeship is a command-line tool and open ecosystem for running AI agents on real-world tasks. A user submits a task, an apprentice agent such as Codex, Claude Code, Cursor, OpenCode or Hermes Agent carries it out in a loop, and a mentor model or a human reviewer evaluates each iteration. The loop continues up to a configurable maximum depth.

Each completed run produces an Experience Compilation: a bundle of the task, the execution traces, lessons and outcomes. Users can inspect and validate these bundles, export them, or publish them to a shared ecosystem. Others can search for them and install the resulting Runtime Training into their own agents, so later runs start from prior work.

The project supports three apprenticeship modes (Autonomous, Expert-Led and Organization Custom) and a Public Ecosystem or Private Internal Only setting. A seed dataset on Hugging Face accompanies the project, containing curated tasks, lessons and execution traces.

Key features

  • 01Runs local agents such as Codex, Cursor, Claude Code, OpenCode and OpenClaw through a unified command.
  • 02Evaluates each loop iteration with a mentor model provider or a human expert.
  • 03Packages completed work into Experience Compilations that can be inspected and validated.
  • 04Exports and installs Runtime Training from prior experience into future runs.
  • 05Supports custom agents through a command template.

Use cases

  • Running repeatable multi-step tasks such as writing a market map or release checklist with an agent that improves across iterations.
  • Sharing agent work experience within a team using a private internal mode.
  • Building a dataset of agent traces and lessons for research on agent training.

Quick start

  1. 1

    Install the CLI globally with npm

    npm install -g agent-apprenticeship
  2. 2

    Run guided setup

    apprentice init
  3. 3

    Check your setup

    apprentice doctor
  4. 4

    Run your first task

    apprentice run "Create a short market map for AI procurement tools."

Things to know

  • Mentor model providers need their own API keys, which you must store locally.
  • It needs a supported local agent CLI installed, such as Codex, Claude Code, or Cursor, or a custom command.

In collections

Open-source AI agent frameworks & appsAI coding assistants & agentsCLI & terminal toolsReinforcement learning frameworks & training environmentsTesting frameworks & test automation tools

Topics

#ai-agents#agentic-ai#cli#agent-traces#reinforcement-learning#coding-agents#llm-evaluation#agent-apprenticeship#agent-economy#agent-experience

FAQ

What is agent-apprenticeship?
A CLI that runs AI coding agents through iterative task loops and turns their completed work into reusable experience data for future agents. Agent Apprenticeship is a command-line tool and open ecosystem for running AI agents on real-world tasks. A user submits a task, an apprentice agent such as Codex, Claude Code, Cursor, OpenCode or Hermes Agent carries it out in a loop, and a mentor model or a human reviewer evaluates each iteration. The loop continues up to a configurable maximum depth.
Is agent-apprenticeship free to use? What license does it have?
agent-apprenticeship is open source under the MIT license. Check the license terms before using it in commercial products.
Is agent-apprenticeship still maintained?
Actively maintained. The last commit to agent-apprenticeship was on Oct 21, 2026.
What can I use agent-apprenticeship for?
Running repeatable multi-step tasks such as writing a market map or release checklist with an agent that improves across iterations. Sharing agent work experience within a team using a private internal mode. Building a dataset of agent traces and lessons for research on agent training.
How do I install agent-apprenticeship?
Install the CLI globally with npm: npm install -g agent-apprenticeship Run guided setup: apprentice init Check your setup: apprentice doctor Run your first task: apprentice run "Create a short market map for AI procurement tools."
Where can I get agent-apprenticeship?
The source code and installation instructions are on GitHub: https://github.com/ray-r-ren/agent-apprenticeship

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects