Docker Open-Sources YAML AI Agent Builder With MCP and RAG

Docker has open-sourced docker-agent, a CLI plugin that lets developers define AI agents in a YAML file and run them with the same command grammar as containers. The repository pitches it as a way to "Build, run, and share AI agents with a declarative YAML config, rich tool ecosystem, and multi-agent orchestration," and the project is Apache-2.0 licensed.

The plugin invokes through `docker agent run` and can pull a definition straight from an OCI registry: `docker agent run myorg/agent:tag`. A third-party walkthrough at runthisai describes it as "a Docker CLI plugin by Docker Engineering that turns the Docker workflow you already know into an AI agent platform," with the agent spec being "a plain `agent.yaml` file that you can version, review, and share like any other infrastructure-as-code artifact."

Model coverage is wide. The README lists OpenAI, Anthropic, Gemini, AWS Bedrock, Mistral, xAI, and Docker Model Runner for local inference. Tool access flows through MCP — local, remote, or Docker-based servers — alongside what Docker calls "Built-in think, todo, and memory tools." Retrieval is handled with BM25, embeddings, hybrid search, and reranking.

It ships as a pre-installed plugin inside Docker Desktop 4.63+, or via Homebrew. The release train is moving fast: GitHub lists v1.149.0 as the current build. Docker's framing — "Create teams of specialized agents that delegate tasks automatically" — puts the project squarely in the crowded agent-builder space, next to the 440 agent stories and 323 open-source items we have tracked in the last 90 days.