OneDev - The powerful and simple DevOps platform

OneDev

Star

Home

Docs

Support

Blogs

Pricing

Sign In

OneDev 16 - the AI autonomous development platform

AI coding becomes much more useful when it is not a side chat. OneDev puts AI users inside the same development system your team already uses for requirements, implementation, review, and delivery.

In OneDev, an AI user is a virtual teammate working from issues, pull requests, builds, code, and workspaces. It can implement assigned issues, open pull requests, respond to review feedback, review other changes, and follow up when CI/CD fails.

The platform difference is where the work starts and where it remains visible. Requirements stay in issues. Execution happens in controlled workspaces. Review happens in pull requests. Validation stays in CI/CD. The AI user participates in that loop instead of replacing it with an external prompt-and-response conversation.

Issues as the Source of Truth

Autonomous development needs a durable work specification. In OneDev, that specification is the issue, or ticket.

An issue captures the feature request, bug report, acceptance criteria, screenshots, design files, documents, and follow-up discussion. The conversation that clarifies the requirement happens where the team can see it, not in a private prompt window. When the requirement changes, the issue changes. When someone needs to understand why the work was done, the issue is the record.

This makes the issue the natural handoff point for an AI user. Instead of writing a fresh instruction from memory, assign the issue or mention the AI user in the issue discussion:

The AI user reads the issue as its primary work specification. Attached screenshots and documents become part of the context. Comments explain edge cases, trade-offs, and decisions. The same issue later guides review, so implementation and acceptance are judged against one shared source of truth.

From Requirement to Running Environment

After the issue is assigned, the AI user can create a workspace, inspect the repository, create an issue branch, make changes, and open a pull request.

This is where OneDev infrastructure matters. The workspace can be created from a spec prepared by the project or organization, with the right container, tools, repository checkout, authentication data, and automation settings. The AI user starts from a known working environment instead of asking every requester to describe setup steps. It is also safer because work runs in an isolated workspace with controlled permissions and entitlement rules.

The pull request keeps the work in the normal development trail: branch, commits, linked issue, review comments, CI/CD checks, and merge decision. A human reviewer does not need to reconstruct what the AI was asked to do. The issue is still there, attached to the change, carrying the original request and the discussion that refined it.

Autonomous Does Not Mean Ad Hoc

OneDev can route AI work by rule, so autonomy does not have to start with a manual mention. Issue field settings can assign matching issues to an AI user, allowing certain product areas, issue types, or priorities to go directly to the agent best suited for them.

The same idea applies to pull requests. Branch protection rules can require selected AI users to review certain changes. Project pull request settings can assign pull requests to AI users, letting them run a final review and merge automatically when all required reviews and CI/CD checks are satisfied.

This turns AI participation into project policy. Teams define where each AI user is allowed and expected to help, and OneDev applies those rules consistently.

Pull Requests Become the Engineering Loop

Review should judge the implementation against the requirement, and OneDev keeps that guidance close. When an AI user submits a pull request from an issue, the linked issue remains the reference for human reviewers and AI reviewers.

Add an AI user as a reviewer to get another review pass. It can inspect the change, compare it with the issue context, leave line-anchored comments, post a review summary, approve, or request changes.

If changes are requested, or if CI/CD fails, the pull request becomes the engineering loop. The AI user continues from review comments and build results, improves the pull request with follow-up commits, resolves addressed comments, and lets the checks run again. The loop stops only when the implementation satisfies the issue, reviewers are satisfied, and required CI/CD checks pass.

This is important for accountability. The AI user is not only using the issue to start work; the team also uses the issue to decide whether the result is correct.

The Platform Shape

The practical shape of autonomous development in OneDev is:

Development needHow OneDev handles it

Define the requirementCapture it in an issue with text, attachments, and discussion<br>Route the workAssign issues and pull request responsibilities manually or by...