What it does
5 features
Docker Agent · Agent orchestration
Product and access
Formerly cagent; a framework for custom agent teams. Docker Gordon is a separate offering.
Core features
5 features from provider sources · checked on 19 October 2026. Availability and limitations follow below.
Usable AI models
Open choice
You set the model per agent in the YAML file and supply your own API key, for example for OpenAI, Anthropic, Gemini, Mistral or xAI. OpenRouter is supported as well. Models run locally through Docker Model Runner, Nativ or any OpenAI-compatible server such as Ollama and vLLM, with no key needed.
Model families named
- OpenAI (GPT)
- Anthropic (Claude)
- Google (Gemini)
- Mistral
- xAI (Grok)
- DeepSeek
- Other models can be connected freely
- Model choice
- You choose
- Your own API key
- Yes
- Router or custom endpoint
- Yes
- Local models
- Yes
The rating covers the generally available main access. Documented · Sources accessed on
Does it fit my task?
2 practical tasks · 3 requirements and limits
A good fit if …
- you want to define agent teams as a YAML file and run them in the terminal
- you want to choose safety modes and permissions for tool calls yourself
- you want to keep sessions locally in SQLite and resume them
Less suitable if …
- you expect a hosted interface with accounts, roles and admin management
- you need ready-made schedules and failure notifications without building them yourself
Editorial assessment based on provider documentation. Not a product test.
Practical tasks
Editorial examples based on documented features. No product tests performed.
Example task 1
- Your input
- Split bug diagnosis and repair between two agents.
- Possible result
- A draft for expert review.
- What to check
- Check inputs, output and performed actions.
Example task 2
- Your input
- Have a second agent review documentation.
- Possible result
- A draft for expert review.
- What to check
- Check inputs, output and performed actions.
Split bug diagnosis and repair between two agents.
Your possible prompt
Design a process for new requests: capture information, sort it by topic, and create a proposed response. Plan an approval step before external actions.
Adjust the task, data and permissions.
Requirements and limits
Requirement
Setup
Install standalone or use Docker Desktop 4.63 or later; configure model access.
Source 1Limitation
Product boundaries
Configure and restrict tool permissions and model access yourself.
Source 1Not yet confirmed
Output quality
Success rates and German-language operation have not been tested here.
Source 1
Can I use it?
With requirements
Europe & regional availability
Installation instructions and model prerequisites reviewed. The assessment covers the local operating path described; complete country and feature availability of all connected services remains open.
Reviewed scope: Local software and documented self-hosting; external services assessed separately
Reviewed on:
With requirements
Self-hosting with own model access
Docker Agent can be installed as a standalone binary or Docker CLI plugin. The quickstart requires credentials for the selected model provider; check its regional availability and optional tools separately.
Region, reason and source
Region covered: Europe: own computer or server; model services are country-specific
Documented reason: Technical assessment based on installation and provider selection. These sources do not establish a legal cause for regional restrictions.
Accounts & permissions
Based on provider documentation. Permissions depend on the account, integration and organization policies. Review the permissions actually requested before use.
Configuration covered: Docker Agent (open source, Apache-2.0) in the terminal or as a local API server on your own machine
Sources reviewed:
Show all 3 points
Open in reviewed sources
Admin approval
The CLI documentation mentions no central approval or restriction by an organisation admin. You configure restrictions yourself through safety modes and permissions.
EvidenceDocumented
Data access & team sharing
Agent configurations are YAML files; according to the docs each agent has its own context and shares no knowledge, and can use built-in tools such as todo lists, memory and task delegation. Configurations are packaged as OCI artifacts. The API server requires an auth token on non-loopback TCP listeners.
Evidence · Evidence
Memory and conversation continuity
5 of 6 criteria documented
Configurable SQLite storage with targeted search and full CRUD tools.
Docker Agent with memory toolset enabled
Sources accessed:
Documentation by criterion
StorageDocumented
Storage and scope
The memory tool stores entries in SQLite; the default is ~/.cagent/memory/<config-name>/memory.db, with a configurable path.
SelectionPartly documented
Selection and timing
The agent writes through add_memory and update_memory. Configuration and agent instructions determine usage; no universal extraction schedule is documented.
Edit & deleteDocumented
Inspect, edit and delete
get_memories, update_memory and delete_memory provide reading, editing and targeted deletion by ID.
Access boundariesDocumented
Agent and channel boundaries
Agents in the same configuration share the database by default. Separate paths per toolset isolate their memories.
RetrievalDocumented
Retrieval
search_memories searches by keywords and/or categories; get_memories retrieves all entries.
New conversationsDocumented
Session reset and compaction
Memory survives sessions. /compact summarizes conversation history; stored sessions and memory are distinct stores.
Cost and effort
Prices as of 19 October 2026
Prices and included usage
Pricing checked
Apache-2.0 software; models and operations are separate, with Docker Desktop licensing where applicable.
- Billing depends on plan, usage and setup.
Setup
Install standalone or use Docker Desktop 4.63 or later; configure model access.
Operation & oversight
Based on public provider documentation. Assessments are editorial inferences from the stated properties. Actual time requirements and reliability have not been measured.
Deployment covered: Docker Agent as a CLI (interactive or --exec) and as an API server on your own infrastructure
Sources reviewed:
Editorial assessment
Sessions, safety modes and resumption are documented, but you build operation, scheduling and failure alerts around the tool yourself. For unattended runs you need to choose the safety mode deliberately.
Show all 6 points
Open in reviewed sources
Sign-ins
Access is via environment variables or tokens; the reviewed pages say nothing about expiry or renewal.
EvidenceDocumented with conditions
Human intervention
Safety modes range from strict to autonomous; restricted fails closed for unattended runs. The docs advise configuring permissions or a safety mode for unattended use. With --yolo tool calls are approved automatically unless explicitly denied. The page does not state the default behaviour without configuration.
EvidenceDocumented with conditions
Detecting errors
--debug writes a log file (by default ~/.cagent/cagent.debug.log), --json emits newline-delimited events (with --exec), and docker agent doctor exits non-zero when an issue would prevent an agent from running. The CLI docs mention no notifications for failed runs.
Evidence
Evidence and open questions
2 sources · 1 open point
Open points
The reviewed sources contain no reliable information on these points. A gap says nothing about the product.
Requirements and limits: not yet confirmed
Sources and evidence
Provider sources reviewed; examples are editorial suggestions. No hands-on test.
Editorial team: Michael Meierhoff · sources reviewed on . Provider information is not independent evidence of performance.
Provider
Product and provider are documented through the linked original sources.
- Operator
- Docker
- Organization
- Provider
- Website
- docs.docker.com
Changes to this profile 1
Added Docker Agent with capabilities, access, pricing and sources.
These notes describe editorial changes. Source and pricing review dates appear in their respective sections.