Our take
Inactiveno commits in over a year- Activity
- 20
- Community
- 24
- Issues
- 21
- Pull requests
- 21
+2 stars in the last 17 days
- Fewer contributors than 82% of the projects we track
This is a useful hands-on blueprint for browser-based Python agent experiments, but its inactive maintenance and concentrated contributor base make it a less dependable foundation for production adoption.
Good fit if
- You want to understand the browser-side experience of running the OpenAI Agents Python SDK through Pyodide.
- You can provide an OpenAI API key or operate an OpenAI-compatible model server, and can work around browser CORS restrictions when using webpage retrieval.
- You are evaluating a demonstrator under the Apache-2.0 license rather than relying on active upstream development.
Look elsewhere if
- You need an actively maintained project: there have been no commits in over a year and no releases in the past year.
- You need evidence of broad maintainer capacity: most of the code comes from one person among three contributors.
- Your use case requires the agent and model inference to run without an external model service.
All health signals
| Last commit | 2025-09-17 (1.1 years ago) |
|---|---|
| Commits, last 90 days | 0 |
| Releases, last 12 months | 0 |
| Contributors | 3 (top contributor: 86% of commits) |
| Issues closed, last 90 days | 0 |
| Pull requests merged, last 90 days | 0 |
| Project age | 16 months |
Checked on 2026-10-10 with the GitHub API.
Overview
Wasm Agents Blueprint demonstrates how to run Python-based AI agents in a browser by combining the OpenAI Agents Python SDK with Pyodide and WebAssembly. It aims to make agent experiments easy to try without setting up an application server or container.
The browser runs the agent code, but inference still comes from an external model service. That can be an OpenAI API or an OpenAI-compatible server such as Ollama or LM Studio, including a locally hosted model.
Key Features
- Runs Python agent code in the browser using Pyodide and WebAssembly.
- Includes a basic conversational agent and a multi-agent handoff example.
- Demonstrates tool calling with webpage retrieval and character counting tools.
- Offers a local-model example that connects to Ollama or LM Studio.
- Uses HTML demos that can be opened directly in a browser.
- Supports OpenAI-compatible model servers as well as OpenAI models.
Use Cases
- Python agent developers can use the demos to explore how an agent SDK behaves in a browser.
- Teams evaluating browser-based AI interactions can prototype without deploying an application server.
- Developers experimenting with local models can try tool-calling agents connected to Ollama or LM Studio.
- Educators and learners can compare a basic agent, handoffs, and tool use through separate examples.
What you need
Detected in the repository
- Python (from pyproject.toml)
- Automated checks on GitHub Actions
License in plain words
Apache-2.0permissive
- Commercial use: yes
- Modify and redistribute: yes
- You must keep: the license, the NOTICE file and a note of your changes
- Share your changes: no
- Includes an explicit patent grant from the contributors:
A summary, not legal advice: the LICENSE file is what applies.
Getting Started
git clone https://github.com/mozilla-ai/wasm-agents-blueprint.git
cd wasm-agents-blueprint/demos
Configure config.js with an OpenAI API key for the OpenAI-backed examples, or start a local model server for local_model.html. Open one of the demo HTML files in a browser. See the README for setup details and troubleshooting.
Considerations
The repository is marked inactive, with no commits in over a year and no recent releases. CI is present, but the repository metadata reports no tests. Contributor activity is concentrated, so maintenance and issue follow-through may depend heavily on one person; there have also been no issues closed in the recent measurement period.
Initial Pyodide package downloads require an internet connection, even when using a local model. Browser CORS restrictions can also prevent the webpage tool from fetching content, and the README suggests disabling those restrictions or using extensions, which should be considered carefully. Model inference remains external to the browser. The project is licensed under Apache-2.0, so review its license terms for your intended distribution and use.
Found this useful?
Share it with someone who would like wasm-agents-blueprint.