Developer framework

Docker Agent

Define a debugging team in YAML and run it from your terminal with your chosen models.

docs.docker.comSources reviewed2026-10-08

Overview

Choose Docker Agent when you want a repeatable, configurable debugging workflow: one agent investigates an error and delegates a proposed fix to another. Keep the YAML, input and test results together so a teammate can review the process. This is a runtime for custom agents, not a hosted website builder or a new model.

Best forDevelopers who want configurable investigation and fix roles for repeatable repository work.

What you provide

A YAML agent configuration, a repository and a reproducible error
Authorized model access or a compatible local model setup

What you get

Agent responses, tool output and proposed code changes
A reusable agent configuration that can be shared through an OCI registry

Fix a website

Setup & workflow

Use Docker Desktop 4.63+ or the stable documentation to install the standalone CLI; earlier Desktop versions used the name cagent.
Connect your selected provider or local model and confirm its cost and data handling.
Adapt the official debugging YAML to a disposable copy of a repository; limit filesystem and shell access.
Run docker agent run debugger.yaml, describe one error, then inspect the proposed diff and rerun tests.

Requirements & installation

Docker Desktop 4.63 or later, or a standalone Docker Agent installation
A supported model connection and explicit access to the project and tools

Limits & review

Configuring multiple agents does not guarantee better results or lower model spend; start with one reproducible issue.
Filesystem and shell tools can change your working environment. Check the actual execution and isolation settings before granting access.
The GitHub Pages documentation tracks the main branch and may describe unreleased features. Match instructions to your installed release.
Official-source review only; Agenlify has not run or benchmarked this product.

Your part

Choose models, review their billing and set permitted tools before running.
Inspect file changes and command output, then run the repository tests yourself.

When to consider another product

A ready-made Docker support chatbot, a no-setup site builder or guaranteed autonomous production repairs.

Tasks & use cases1

Fix bugs in an existing project

The stable documentation includes an investigator-and-fixer workflow for repository bugs.

Prepare: A YAML agent configuration, a repository and a reproducible error · Authorized model access or a compatible local model setup
Get: Agent responses, tool output and proposed code changes · A reusable agent configuration that can be shared through an OCI registry
Your part: Choose models, review their billing and set permitted tools before running. · Inspect file changes and command output, then run the repository tests yourself.

Plans & billing details

Cost planning

This review verifies the runtime license and separate cost boundaries, not numeric provider rates or Docker subscription allowances.
Using a local model avoids that cloud API path but does not remove hardware, electricity or model-license requirements.

Published prices are a snapshot. Confirm billing cycle, taxes and current allowances with the provider.

Sources & verification3

This profile is based on official sources, not a hands-on product test.

Vendor descriptions and demos document advertised features. Editorial guidance is based on these sources.

Docker Agent stable documentationdocs.docker.com
Documents YAML agent teams, installation, model credentials, debugging and OCI sharing; distinguishes Docker Agent from Gordon.
Docker Agent official repositorygithub.com
Documents the CLI, cloud model providers and the local Docker Model Runner option.
Docker Agent Apache-2.0 licensegithub.com
The runtime has an Apache-2.0 license; model services and other Docker products have separate terms.

Frequently asked questions

Is Docker Agent the same as Gordon?

No. Docker Agent runs custom agent teams from configuration files. Gordon is Docker’s built-in AI assistant, accessed with docker ai.

Is Docker Agent free, including model usage?

Free Apache-2.0 runtime. Cloud model API usage is billed separately by your provider; local models require your own compute. Docker Desktop, registries and other Docker services have separate terms and possible charges.

Does the Docker name guarantee my agent is isolated?

Do not infer isolation from the product name. Review the tool permissions, execution environment and the sandbox instructions for your installed release before running repository changes.