[Agent]: Add support for Docker Agent

[nervgh]

Agent Name

Docker Agent

Official Website

https://docs.docker.com/ai/docker-agent/

Agent Type

CLI tool (command-line interface)

CLI Command (if applicable)

docker agent or docker-agent

Installation Method

Docker Desktop (Pre-installed); brew install docker-agent; Download Binary Release

Command/Workflow Structure

Docker Agent uses YAML configuration files to define agents, their instructions, models, tools, sub-agents, and hooks.

A typical configuration is started with:

docker agent run ./agent.yaml

Spec Kit skills are installed in the project-local directory:

.docker-agent/skills/

Each skill follows the Agent Skills layout:

.docker-agent/skills/speckit-<command>/SKILL.md

For example:

.docker-agent/skills/speckit-specify/SKILL.md
.docker-agent/skills/speckit-plan/SKILL.md
.docker-agent/skills/speckit-tasks/SKILL.md
.docker-agent/skills/speckit-implement/SKILL.md

Argument Passing Pattern

Docker Agent accepts the user message as a positional argument:

docker agent run ./agent.yaml "Read the Spec Kit instructions and create a specification"

For non-interactive execution, it supports:

docker agent run ./agent.yaml --exec "Read the Spec Kit instructions and create a specification"

For JSON output:

docker agent run ./agent.yaml --exec --json "Read the Spec Kit instructions and create a specification"

The resulting command includes the configured agent file:

docker agent run --exec --json ./agent.yaml "<prompt>"

Popularity/Usage

Growing adoption (hundreds of users)

Documentation Links

• Docker Agent overview:
  https://docs.docker.com/ai/docker-agent/
• Installation:
  https://docs.docker.com/ai/docker-agent/getting-started/installation/
• Quick start:
  https://docs.docker.com/ai/docker-agent/getting-started/quickstart/
• Agent configuration:
  https://docs.docker.com/ai/docker-agent/configuration/overview/
• Hooks:
  https://docs.docker.com/ai/docker-agent/configuration/hooks/
• Skills:
  https://docs.docker.com/ai/docker-agent/features/skills/
• Docker Agent repository:
  https://github.com/docker/docker-agent

Use Case

Manual usage

Docker Agent provides a Docker-native way to run AI coding agents defined through version-controlled YAML configurations. Supporting it in Spec Kit allows users to install and use Spec Kit skills from .docker-agent/skills/ while preserving Docker Agent’s existing configuration and execution model.

The integration supports the standard Spec Kit workflow:

specify → plan → tasks → implement

Docker Agent can be launched from the project root and can access the repository’s .specify/ directory, generated specifications, plans, and task files.

Workflow automation usage

The workflow would start with:

specify workflow run speckit \
  -i spec="Add a user authentication system" \
  -i integration=docker-agent

Workflow automation usage with per-step configuration

See more details about of this type of usage in related issue #4275

The workflow would start with:

specify workflow run speckit \
  -i spec="Add a user authentication system" \
  -i integration=docker-agent

and configuration may look like this:

steps:
  - id: specify
    command: speckit.specify
    integration: docker-agent
    integration_args: [ ./requirements-agent.yaml ]  # proposed changes
    # integration_options: some opts                 # proposed changes
    input:
      args: "{{ inputs.spec }}"

  - id: plan
    command: speckit.plan
    integration: docker-agent
    integration_args: [ ./planning-agent.yaml ]   # proposed changes
    # integration_options: some opts              # proposed changes

    input:
      args: "{{ inputs.spec }}"

  - id: implement
    command: speckit.implement
    integration: docker-agent
    integration_args: [ ./implementation-agent.yaml ]  # proposed changes
    # integration_options: some opts                   # proposed changes
    input:
      args: "{{ inputs.spec }}"

The corresponding executions would use different files:

docker agent run ./requirements-agent.yaml --exec --json "<specify prompt>"
docker agent run ./planning-agent.yaml --exec --json "<plan prompt>"
docker agent run ./implementation-agent.yaml --exec --json "<implement prompt>"

Example Command File

Contribution

• I can help test the integration
• I can provide example command files
• I can help with documentation
• I can submit a pull request for the integration

Additional Context

No response