Practical overview
When it fits
Use it when you want an example project to inspect or adapt for an agent workflow, especially when Docker Compose is part of your development setup.
Who it is for
- Developers learning or comparing agent frameworks through runnable Compose examples.
- Teams prototyping self-hosted agent workflows with local open models or remote model providers.
Documented uses
- Find a Compose-based starting point for a particular agent workflow.
- Explore combinations of agent systems, models and MCP-connected tools.
- Launch a demo locally or through a cloud context using its project configuration.
Requirements & limits
The repository is public. Docker contributions are dual-licensed under Apache-2.0 or MIT; individual examples may have separate licenses.
- Local open-model execution requires GPU-capable hardware and platform-specific Docker Model Runner requirements; Docker Offload is listed as an alternative to a local GPU.
- Linux Docker Engine requires Docker Compose 2.38.1 or later.
- Some demos require MCP tokens; the README instructs users to provide them through .mcp.env where an example environment file exists.
- OpenAI use requires an API key file and a demo's OpenAI Compose configuration.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Start a documented Compose demo
Configure and start one of the repository's self-contained agent demos.
Before you start
- Docker Desktop 4.43.0 or later or Docker Engine.
- For local open models, a GPU-capable laptop or workstation; Docker Offload is an alternative.
- For Docker Engine on Linux, Docker Compose 2.38.1 or later. Linux Engine and Windows Desktop users must meet Docker Model Runner GPU requirements and have the necessary drivers for that setup.
Steps
Change to the root directory of the chosen demo project.
If the project includes mcp.env.example, create .mcp.env from it and supply the required MCP tokens. If it does not, the demo does not need secrets.
Run docker compose up --build.
Expected outcome
The selected demo is started using its Compose configuration.
Guide limitations
- Requirements vary with the Docker platform and whether models run locally.
- Individual examples may have their own license terms.
Start a demo with OpenAI models
Use the documented OpenAI configuration instead of running models locally with Docker Model Runner.
Before you start
- A demo with compose.yaml and compose.openai.yaml configurations.
- An OpenAI API key.
- Docker Compose.
Steps
Create a secret.openai-api-key file containing the OpenAI API key.
Start the project with docker compose -f compose.yaml -f compose.openai.yaml up.
Expected outcome
The demo starts using its OpenAI Compose configuration.
Guide limitations
- The README describes this option for demos generally; model and tool configuration may differ by example.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 12 October 2026
- Original website title
- docker/compose-for-agents · Official homepage title · retained in its source language