agent-apprenticeship
The living ecosystem where AI agents complete tasks through workflow loops, improve through iterative execution, are evaluated by mentor agents or humans in the loop, and turn completed work into reusable work experience and data to improve future agents.
What is it?
What it is
Agent Apprenticeship is an open-source ecosystem enabling AI agents to learn through real-world task execution in iterative workflow loops. Agents complete tasks with mentor guidance, generate reusable learning signals, and share experiences to improve collective agent performance.
Why it exists
To create a compounding exchange of agent work experience where real-world task execution generates training signals, which are reused to enhance future agent capabilities and economic value across domains.
Who should use it
AI engineers building autonomous agent workflowsDevelopers working with agentic frameworks like Claude Code, Cursor, or OpenClawMachine learning researchers focused on reinforcement learning and agentic training signalsOrganizations looking to build private, internal agent learning ecosystemsData scientists seeking high-quality agent execution traces and experience compilations
Who should avoid it
Users looking for a simple chatbot interfaceDevelopers who do not work with CLI-based agent toolsIndividuals without access to LLM API keys (OpenAI, Anthropic, etc.)
How it works
A quick walkthrough in plain English
How agent-apprenticeship works
Step 1 of 3
You interact with it
Open agent-apprenticeship, send a request, or connect it to your stack.
Features
Advantages
- Open-source and MIT licensed, encouraging community contributions
- Extensive pre-built seed dataset accelerates onboarding
- Modular design allows easy addition of new agents or model providers
- Reusable training signals enable compounding learning across tasks
- Economic value estimation provides insight into agent productivity
- Flexible configuration (CLI, env vars, .env.local) simplifies setup
- Supports both autonomous and expert-led workflows
- Public ecosystem mode promotes sharing and collaboration
- Runtime training installation streamlines reuse of learned skills
- Clear documentation and quickstart guide reduce learning curve
Disadvantages
- Requires API keys for external model providers, adding setup complexity
- Limited to agents with CLI support; custom agents need manual configuration
- Learning curve for configuring loops, experience compilations, and economic estimation
- Performance overhead from iterative loops and data handling
- Seed dataset may not cover all specialized domains
- Potential security concerns with public ecosystem mode
- Requires Node.js/npm environment, which may be unfamiliar to some Python users
- Documentation may need updates for newer agent integrations
Installation
FAQ
How do I install Agent Apprenticeship on my machine?
You can install it locally with the one‑liner `npx agent-apprenticeship init`. If you prefer a global install, run `npm install -g agent-apprenticeship` and then use the `apprentice` command. The `init` step will set up the necessary configuration files.
How do I run a task with Agent Apprenticeship?
After initialization, execute a task with `apprentice run "<task description>"`. For example: `apprentice run "Create a short market map for AI procurement tools."`. The system will launch the chosen apprentice agent, run the workflow loop, and output the run folder and experience compilation path.
Where do I store my model provider API keys?
Create a file at `~/.agent-apprenticeship/.env.local` and add your keys, e.g. `OPENAI_API_KEY="..."`, `ANTHROPIC_API_KEY="..."`, etc. Alternatively, export them in your shell session with `export OPENAI_API_KEY="..."` before running commands.
What apprenticeship modes are available and how do I choose one?
During configuration, you can select from `Autonomous`, `Expert‑Led`, or `Organization Custom`. Use `apprentice configure` (or `apprentice configure model`) to set the mode. The mode determines whether the agent runs autonomously, is guided by a mentor agent, or follows a custom workflow.
How can I inspect or export the experience compilation from a run?
After a run completes, use `apprentice ecosystem inspect <experience_compilation_path>` to view details. To verify the bundle, run `apprentice bundle check <experience_compilation_path>`. For a full export, use `apprentice ecosystem export <experience_compilation_path> --full`.
How do I contribute or pull data from the public ecosystem?
Set the contribution mode to `Public Ecosystem` via `apprentice settings`. Then you can search with `apprentice ecosystem search <query>`, inspect with `apprentice ecosystem inspect <id>`, pull data with `apprentice ecosystem pull <id>`, and export or install runtime training from the retrieved experience compilations.
Featured in Videos
YouTube tutorials and walkthroughs for agent-apprenticeship
Alternatives
Similar projects ranked by category, topics, and text overlap.
