Requirement and task management
Organize bounded AI work around requirements, projects, and task history rather than isolated prompts.
PRODUCT FEATURES · FACTS CHECKED 2026-09-08
Features around the agent, not only inside the prompt.
MonkeyCode is positioned as an AI development platform for engineering teams. Its product thesis combines requirements, tasks, environments, models, projects, desktop and mobile clients, and deployment choices into one managed workflow.
The MonkeyCode task workspace
Original product image with English text. Interface and account values vary by version.
Documented capability map
Eight layers to evaluate.
These descriptions summarize public project materials; they are not performance, security, or compatibility guarantees.
MONKEYCODE / WORKFLOW
From a brief to a reviewable change
The model handles the task. The workspace gives it somewhere to work. You decide what to keep.
Your brief
Define the goal and allowed scope.
Model + workspace
Run the task with files, terminal and tests.
Reviewable change
Inspect the diff and verify the result.
Server-side development environments
Run repository work with files, terminal, build, test, ports, and preview workflows in managed environments.
Desktop client with local agent (MonkeyWork)
Authorize a working directory and let the agent read, create, and modify local code, documents, and assets; a paired browser extension assists with web tasks.
Mobile follow-up
View task progress, continue conversations, and receive results on Android and iOS while tasks keep running.
Model management
Select from model families exposed by the current platform configuration; exact versions and routes must be verified.
Team visibility
Keep project and task context visible beyond one developer session.
Hosted or private deployment
Evaluate the hosted service or operate a documented private/offline deployment path.
Open-source codebase
Inspect and modify the public AGPL-3.0 codebase subject to its license terms.
Product tour
The managed workflow, in the official product images.
Official project screenshots. Interface details change by release; verify the version you evaluate.
Evaluation workflow
Trace a requirement to accepted evidence.
Public materials establish the components; your pilot must verify how completely they connect in the release you test.
- 01Define a bounded requirement
Write acceptance criteria, repository scope, permissions, and prohibited actions.
- 02Create and execute an AI task
Confirm environment startup, repository access, selected model, commands, and retained history.
- 03Build and test
Keep command output, failures, retries, previews, and artifacts as review evidence.
- 04Review the change
Inspect the diff and acceptance criteria. Verify branch, commit, and PR behavior rather than assuming automation.
Deployment distinction
Hosted and self-hosted move responsibility.
Do not assume identical features, models, integrations, operations, or security controls across deployment modes.
| Area | Hosted evaluation | Self-hosted evaluation |
|---|---|---|
| Platform operation | Provider-operated service | Organization owns installation, capacity, upgrades, and recovery |
| Environment boundary | Verify provider behavior and terms | Verify your hosts, network, isolation, and storage |
| Models and integrations | Confirm currently available configuration | Confirm release support, credentials, routes, and provider terms |
| Security evidence | Request current service evidence | Test configured controls and operational processes |
Common questions
Features, answered.
Related: is it open source?, how it differs from editor tools, plans and pricing.
What can MonkeyCode do?
Documented capabilities include requirement and task management, server-side development environments, model management, team visibility, hosted or private deployment, a desktop client with a local agent (MonkeyWork), mobile follow-up, and an open-source AGPL-3.0 codebase.
Does MonkeyCode do code completion?
No. The hosted web app is not positioned as editor autocomplete; it coordinates bounded AI tasks in managed server-side environments. The desktop client adds a local agent for authorized directories — not an autocomplete layer. Editor-first tools cover in-IDE completion, and some teams use both.
Do MonkeyCode tasks run in the cloud?
The hosted service runs tasks in server-side development environments with build, test, and preview workflows — no local setup. The desktop client can also run a local agent on your machine and continue cloud tasks across devices.
Is MonkeyCode open source?
Yes. The public chaitin/MonkeyCode repository is licensed under GNU AGPL-3.0, so you can inspect and modify it subject to the license terms.
Are hosted, desktop, and self-hosted features identical?
Do not assume so. Where tasks run, local file access, browser pairing, mobile reach, and admin controls differ by mode; see the execution-mode boundaries on the product page and verify each in the mode you plan to use.
VERIFY THE SYSTEM