AgentRail provides a local-first control plane for coding agents, streamlining the entire project lifecycle from issue intake to code shipping. It offers a unified, compact API for managing agent workflows, making it ideal for developers focused on local evaluation and self-hosting of agent-based development processes.

How It Works

AgentRail acts as a central control plane, abstracting the complexities of agent task management. It offers a single API endpoint for issue intake, routing to appropriate agents, managing assigned work, submitting Pull Requests, integrating with CI pipelines, processing review feedback, and handling shipping. The system emphasizes retry-safe mutations using idempotency keys and provides real-time task events via SSE and webhooks, alongside local provider integrations for common development tools.

Quick Start & Requirements

• Prerequisites: Node.js 24 or newer, npm. Optional provider credentials (GitHub, CircleCI, Linear) for live integrations.
• Install CLI: npm install -g @agentrail-core/cli or use npx @agentrail-core/cli init.
• Initialize: Run agentrail init for interactive setup, configuring local state under ~/.agentrail.
• Start Server: agentrail server start.
• Verify Setup: agentrail doctor --agent-id agt_example.
• Documentation: Developer guide, SDK guide, Five-minute quick start, Integration guide.

Highlighted Details

• Local-first, single-instance runtime for self-hosted agent control plane.
• Unified API covers issue intake, routing, PR submission, CI, review, and shipping.
• Retry-safe mutations via idempotency keys and real-time task events (SSE/webhooks).
• Local provider setup for GitHub, CircleCI, and Linear, with configurable PR shipping logic.
• Agent role templates and repository maps for flexible agent configuration, with optional AI-assisted map generation.

Licensing & Compatibility

AgentRail is distributed as source-available software, not open source. The specific terms are detailed in the LICENSE file, which may impose restrictions on commercial use or integration with closed-source projects.

Limitations & Caveats

The current release focuses on a self-managed, single-instance runtime for local evaluation and self-hosting. A managed team/fleet layer, AgentRail Cloud, is planned separately. The "source-available" license requires careful review to understand its implications for adoption, particularly in commercial or proprietary contexts.