jayminwestAI apps
warren
Self-hosted control plane for coding agents—observe, schedule, and manage agent lifecycles like infrastructure.
- Browser
- Self-hostable
- Docker supported
- Needs API key

- Popularity
- 481 Stars
- GitHub stars
- Recent activity
- 10/9/2026
- Updated in the last 30 days
- License
- MIT
- Permissive
Why it matters
We look beyond stars: what problem it solves, whether it creates real utility, and what makes its approach worth noticing.
Problem
AI coding agents often lack isolation and observability when run locally, making stable execution on servers or CI difficult. This project provides a self-hosted control plane for this pain point.
Practical value
Provides single-container deployment and Kubernetes runtime support with sandbox isolation, live event streaming, mid-run steering, and scheduled task dispatch.
Innovation / differentiation
Unifies multiple mainstream coding agents under a single scheduling interface, combining sandbox technology with short-lived run management.
Leverage potential
Enables rapid deployment of the complete infrastructure containing the control plane and sandbox runtime via a single Docker image or Compose file.
Why now
As terminal-based AI coding tools proliferate, team demand for server-side hosting, isolation, and governance of agent execution is growing.
Community activity
In the last 90 days there were 99 new issues and 675 pull requests; the bounded issue/PR samples include 6 issue authors and 5 PR contributors, with at least 20 releases.
Maintainer responsiveness
The 99-issue window sample had a 98% close rate, and the 100-pull-request sample had a 98% merge rate. Maintainer-response observations covered 70% of that issue sample, with a 55% response rate and median first response of 1.8 days.
Key highlights
- One image, one volume. The supervisor (
src/supervisor/main.ts) is the container ENTRYPOINT. It spawns the sandbox - Native sandboxing per run. In the default
localtopology every run gets a freshbwrap-isolated workspace under ` - Built-in agents.
claude-code,sapling, andpiship inline (src/registry/builtins/), so a dispatch needs no e
Quick start
How it is installed, how hard it is, and where to start.
Best for
- Teams that want data on their own servers
- People who prefer Docker deploys
Watch outs
- 5 inferred from materials — review under Sources below
More about it
Warren brands itself as “Coolify for coding agents”: treat agents as observable, schedulable services instead of disposable terminal sessions. The lens is lifecycle and operations.
It fits people already running many agent processes who want a single pane of glass. You trade control-plane complexity for operability; casual solo use may not need it.
Sources
Each field shows its status and source — expand to review.
10 · Expand
Sources
Each field shows its status and source — expand to review.
Latest release
Verifiedv0.19.3
Source: GitHub API · latest_release=v0.19.3 · 10/10/2026
License
VerifiedMIT
Source: GitHub API · license.spdx_id=MIT · 10/10/2026
logo url
VerifiedNone
Source: admin_cms · cms editor · 8/28/2026
needs api key
Inferred from materialsYes
Source: Project README · matched api key keywords · 8/15/2026
One-liner
Verified{"en":"Self-hosted control plane for coding agents—observe, schedule, and manage agent lifecycles like infrastructure.","zh":"Warren 将 AI 编码代理从本地终端会话提升为可观测、可调度、具备生命周期管理的生产级基础设施。它解决了代理运行缺乏隔离、难以监控和无法在 CI/CD 环境中稳定执行的痛点。"}
Source: heuristic_batch_publish · batch-publish-tools heuristic write · 8/24/2026
Platforms
Inferred from materialsbrowser
Source: Project README · browser · 8/15/2026
Category hint
Inferred from materialsai-apps
Source: Project README · hint=ai-apps · 8/15/2026
screenshot url
VerifiedNone
Source: admin_cms · cms editor · 8/28/2026
supports docker
VerifiedYes
Source: Repository file · dockerfile=true; compose=true · 10/10/2026
supports self host
Inferred from materialsYes
Source: Project README · matched self-host keywords · 8/15/2026
Related projects
Other verified projects matched by category, capabilities, and intended roles.
hermes-agent
A personal AI agent that runs locally or on remote infrastructure with terminal, messaging, memory, skills, and scheduled automation.
ollama
A local model runner for Windows, macOS, and Linux with a REST API plus Python and JavaScript clients.
ragflow
An open-source RAG engine fusing deep document understanding and agent capabilities to build grounded context layers for LLMs.
unsloth
A desktop and developer toolkit for running, fine-tuning, and deploying language and diffusion models locally on Windows, macOS, and Linux.
financial-services
Financial services reference agents, skills, and MCP connectors for Claude and Managed Agents.
OpenCLI
Convert websites and logged-in Chrome sessions into CLI interfaces for humans and AI agents.
