Proof of Use. An independent field guide · No. 01
Back to the collection
Coding & Agents

OpenHands

A self-hostable agent workspace whose Agent Canvas connects coding agents and automation workflows to local, container, remote or cloud backends.

coding agentself-hostedcontainerstestingautomation
Source-reviewedNot hands-on testedSources checked

Start with an idea

What you can make

01

Give an agent a bounded bug fix and ask for a regression test.

02

Add tests or a small feature in an existing repository, then review the resulting changes.

Official examples

See the documented workflow

Official prompts and expected workflows cover testing, bug fixes and feature work; these are examples rather than independently verified outcomes.

Official development-task examples

Before you begin

What it takes to run

Backend and workspace setup

  • For a local non-container install, the current README requires Node.js 24 or later and uv.
  • For the documented sandbox route, install Docker and explicitly choose the project directory the agent can access.
  • Configure model access and the desired backend; internet-facing deployments need the self-hosting security guidance.
Runs through / on

Browser-based Agent Canvas · macOS · Linux · Windows · Docker · Remote or cloud backends

The real tradeoffs

Cost & licensing

Running costs

Local open-source software is free; budget for model inference and your compute. Hosted and Enterprise offerings have separate pricing.

The Individual cloud tier still uses your model key or billed provider usage and has service limits. Enterprise pricing is custom; no universal per-task cost is implied.

Code license

MIT (current Agent Canvas repository)

Read the code license

Models & other terms

No inference-model rights are supplied by this code license; selected models and providers have separate terms.

Current root LICENSE is MIT. Official docs classify OpenHands Cloud as commercial SaaS and Enterprise as commercial; do not apply the repository license to the whole product family or older releases.

Know the limitations

  • The non-sandbox local launcher gives the agent direct filesystem access.
  • Sandboxing, orchestration and infrastructure configuration add operational work; separate workspaces avoid conflicting concurrent edits.
  • Cloud and Enterprise features are not identical to the open-source local setup; human code review and tests remain necessary.

Follow the evidence

Sources & verification

Official sources were reviewed on October 14, 2026. We did not install, run or benchmark this project. This is a source-check date, not a release date.

Links and project terms can change. These sources document the claims above; they are not a safety certification. Read our full method.

Keep exploring

A few other possibilities

All tools

01Coding & Agents

Aider

A terminal-based coding assistant that edits an existing repository with a chosen local or hosted language model and tracks changes in Git.

Try it for

Make a coordinated change across several source files, then review the Git diff.

Read entry Official examples
Setup · Moderate
Terminal + model connection
Cost to consider
API usage or local compute

02Coding & Agents

Goose

An extensible desktop and command-line agent that can edit code, run development tools and use MCP integrations with a chosen model provider.

Try it for

Prototype a small browser app from a concrete specification.

Read entry Official examples
Setup · Moderate
Desktop/CLI + model connection
Cost to consider
API/subscription or local compute

03Coding & Agents

Cline

A coding agent for editors, the terminal and desktop that can modify files, run commands and use tools through configurable approval controls.

Try it for

Build and iterate on a small web feature while reviewing proposed file changes.

Read entry Official examples
Setup · Moderate
Editor/app + model connection
Cost to consider
API/subscription or local compute