Build a learning application with LAMB
We develop LAMB on GitHub and use the dev branch for development. You can install LAMB on your own machine by giving an AI coding agent the file Documentation/installation/deployLocal.md: it is written for an agent to follow, and the agent asks you for the values it needs. Everything runs on localhost with Docker.
LAMB provides a management layer for assistants and model access that educational applications can use within an institutional environment. Build a new teaching tool, automate a workflow or contribute to the free-software ecosystem. LAMB is licensed under GNU GPL v3.
Build a tool for a learning activity
The public lamb-tool-template demonstrates a tool launched from an LMS, using an authorized LAMB assistant and returning an instructor-approved result to the LMS. It gives an application team a concrete place to begin.
You need a LAMB installation that supports creator API access, permission to use that access, and a published assistant available to your account. An institution must also permit the LTI version used by the tool. The demonstrated template uses LTI 1.1.
See the illustrated tool walkthrough →
Tools for development and testing
| Tool | What it helps you do | Public source |
|---|---|---|
| LAMB LTI Test Tool | Exercise LTI 1.1 launches, instructor/learner roles, launch parameters and Basic Outcomes grade return during development | LTI Test Tool |
| Moodle CLI | Work with Moodle’s standard web-service APIs from the command line, within the account’s permissions | Moodle CLI |
| LAMB CLI | Manage assistants, knowledge bases, rubrics, tests, analytics and supported agent workflows from the terminal | LAMB CLI |
| LAMB tool template | Start a separately hosted educational application with authorized assistant access and LMS integration | Tool template |
Use the LTI Test Tool to inspect and rehearse an integration, then validate it in the institution’s actual LMS configuration. Its development environment is not a certification of Moodle or Canvas compatibility.
Moodle CLI and LAMB’s Moodle connector use standard web services; they do not require a custom Moodle plugin. The available operations depend on enabled services, account permissions and, for the connector, LAMB’s organization policy.
Understand the public contract
LAMB’s creator API exposes supported models and chat completions through an OpenAI-compatible interface on deployments with that capability. Compatibility describes the supported interface; it does not promise every endpoint or option of another service.
Use the assistant identifiers returned by the API. Keep personal keys on the tool’s server, never in student-facing JavaScript. Check permission and version prerequisites before assuming an empty response means a feature is broken.
The template repository provides its public setup instructions. Pin the revision you test and record the LAMB version it is paired with. The template walkthrough links to a dated revision rather than describing an untested future checkout.
Contribute to the project
LAMB is distributed under GPL v3. You can report an issue, improve documentation, test a change or propose a contribution through the source repository.
Prepare a useful contribution →
If an AI agent is helping you evaluate or develop a tool, start with the public agent reading guide. Public documentation describes supported use; internal architecture and design notes are not part of this website.