open-ace: Self-Hosted Workspace for AI Coding Agents

Overview

Open ACE is a self-hosted workspace, remote execution layer, and governance control plane for AI coding agents. It gives teams a browser-based way to run coding-agent CLIs on their own machines, while administrators manage access, API key proxying, quotas, cost visibility, and audit trails. The project is built with a Python and Flask backend and a React and TypeScript frontend.

It is aimed at organizations bringing AI coding agents into day-to-day engineering, especially those that need remote execution inside their own network and governance across users and teams. Its autonomous development workflows can also connect GitHub issues to task execution and pull-request activity.

Key Features

• Browser workspace for coding-agent sessions, prompts, history, and project context.
• Remote Agent execution on team development, staging, or GPU machines, with terminal and remote directory access.
• Central API key proxy that keeps real keys on the server and issues short-lived, revocable proxy tokens.
• Usage and governance tools for quotas, costs, anomalies, audit trails, compliance reports, and ROI estimates.
• GitHub issue-driven workflows for planning, implementation, pull-request creation, comments, and human pause or resume.
• Multi-tenant access controls and integrations including OIDC/OAuth2/SAML SSO and Feishu or DingTalk organization sync.
• Deployment options using Docker Compose, with Kubernetes manifests also included in the repository.

Use Cases

• An engineering team wants developers to run coding agents on internal machines without distributing long-lived model API keys to those machines.
• An AI platform or IT team needs centralized visibility into usage, quotas, costs, and audit activity across users.
• A team wants to turn GitHub issues into agent-assisted development tasks that can pause for human input and continue through pull-request work.
• An organization needs a self-hosted workspace for remote or GPU-hosted development environments rather than sessions limited to individual laptops.

Getting Started

git clone https://github.com/open-ace/open-ace.git
cd open-ace
docker compose up -d

Open http://localhost:19888. Review the README and Deployment Guide for production configuration, multi-user deployment, and security requirements.

Alternatives

• Vicoa: Vicoa coordinates coding-agent teams across clients and machines, while Open ACE centers on self-hosted execution with centralized key proxying and governance.
• codeg: Codeg focuses on collaborative agent sessions, tasks, and code changes; Open ACE emphasizes remote execution, key management, and audit controls.
• multica: Multica assigns issues to agents on controlled machines and supports human review, while Open ACE adds centralized credential proxying and governance workflows.
• OpenHands: OpenHands runs agents across local, remote, and cloud backends and automates engineering tasks; Open ACE focuses on team workspaces, access controls, and usage audits.

Considerations

• This is a relatively young project. Its recent commit, release, and issue activity indicate active development, but the platform has not yet had a long period to establish operational maturity.
• Production deployment requires explicit security configuration, including strong secrets and changing the documented default credentials. Multi-user mode runs the container as root to create system users, so teams should follow the dedicated overlay and deployment guidance.
• The service uses Python 3.10 or later and supports Docker Compose. Teams should also plan for a database and the remote machines on which agents will execute; the README lists PostgreSQL and SQLite as database options.
• The Apache-2.0 license permits broad use, subject to its license terms. Review those terms and any deployment-specific obligations before adopting it.