Summary

Agent Apprenticeship provides an open infrastructure for AI agents to learn from real-world work. It facilitates iterative workflow loops, enabling agents to generate reusable learning signals and improve performance through collective experience exchange, aiming to unlock economically valuable agent tasks. This project is designed for developers and researchers seeking to enhance AI agent capabilities and foster a collaborative learning environment.

How It Works

The system operates on a compounding exchange model: task execution generates training signals, which enhance future work, thereby creating new reusable experience. Apprentice agents collaborate with mentor agents or human experts on long-horizon tasks. Each workflow loop transforms execution into shared improvement, building a living ecosystem of agent knowledge.

Quick Start & Requirements

Installation is via npx agent-apprenticeship init or global npm install -g agent-apprenticeship followed by apprentice init. Prerequisites include Node.js/npm, and for ecosystem features, the GitHub CLI must be installed and authenticated. API keys for model providers (OpenAI, Anthropic, Gemini, OpenRouter) are required and configured via ~/.agent-apprenticeship/.env.local or environment variables. The system auto-detects compatible Apprentice Agents (e.g., Codex, Cursor, Claude Code) and allows custom agent configuration. Official repository: https://github.com/Forsy-AI/agent-apprenticeship.

Highlighted Details

• Features a seed dataset comprising 500+ curated tasks, 495 reusable lessons, and over 1000 execution traces/episodes.
• Supports multiple Apprentice Agents and Mentor Modes (model-assisted, expert-led, hybrid).
• Enables an open ecosystem for sharing and discovering agent experience packages, with configurable auto-sharing options.
• Includes functionality to estimate task-level economic value for agent work.

Maintenance & Community

The project is hosted on GitHub at https://github.com/Forsy-AI/agent-apprenticeship. Specific details regarding active maintainers, community channels (like Discord/Slack), or sponsorship are not detailed in the provided README.

Licensing & Compatibility

No license information is specified in the provided README content. This absence makes it impossible to assess compatibility for commercial use or closed-source linking without further inquiry.

Limitations & Caveats

The system's full functionality relies on the presence and correct configuration of specific Apprentice Agent CLIs and model provider API keys. Ecosystem sharing features necessitate GitHub CLI installation and authentication. The lack of explicit licensing information is a significant caveat for adoption decisions.