---
title: "Deploy Plandex | Open Source AI Coding Agent for Large Codebases"
description: "Self Host Plandex. AI coding, plan versioning, multi-file edits, & more"
category: "AI/ML"
url: https://railway.com/deploy/plandex
---

# Deploy Plandex | Open Source AI Coding Agent for Large Codebases

Self Host Plandex. AI coding, plan versioning, multi-file edits, & more

**[Deploy Plandex | Open Source AI Coding Agent for Large Codebases on Railway](https://railway.com/template/plandex)**

- **Creator:** Heimdall
- **Category:** AI/ML
- **Total deploys:** 8

## Template content

### Postgres https://devicons.railway.app/i/postgresql.svg

- **Image:** ghcr.io/railwayapp-templates/postgres-ssl:18

### Plandex Server https://raw.githubusercontent.com/plandex-ai/plandex/main/images/plandex-logo-thumb.png 

- **Image:** plandexai/plandex-server:latest
- **Public domain:** Yes

## Documentation

![Plandex logo](https://docs.plandex.ai/img/plandex-social-preview.png)

# Deploy and Host Plandex on Railway

Plandex is an open-source, terminal-based AI coding agent built for large projects and real-world development tasks. It plans multi-step changes across dozens of files, applies them in a sandboxed diff review, and supports version control and rollback for every AI-generated plan.

Deploy Plandex on Railway with a pre-configured PostgreSQL database, persistent volume for plan storage, and an embedded LiteLLM proxy that routes requests to 11+ LLM providers. Self-host Plandex to keep your code and API keys on your own infrastructure while your team collaborates on AI-assisted development.

## Getting Started with Plandex on Railway

After deployment completes, install the Plandex CLI on your local machine. The following script installs the CLI for your platform:

```
curl -sL https://plandex.ai/install.sh | bash
```

Set your Plandex server URL and sign in:

```
export PLANDEX_API_URL=https://your-railway-domain.up.railway.app
plandex sign-in
```

Create a new account when prompted — in development mode with LOCAL_MODE enabled, email verification is skipped. Navigate to your project directory, run `plandex new` to start a plan, then `plandex tell "implement feature X"` to begin coding with AI. Set your LLM API key locally with `export OPENAI_API_KEY=sk-...` or `export ANTHROPIC_API_KEY=sk-ant-...` before starting.

## About Hosting Plandex

Plandex is a client-server AI coding agent where the CLI runs on developer machines and connects to a central server. The server manages plans, file context, conversation history, and model routing through an embedded LiteLLM proxy.

- Plans changes across 50+ files with cumulative diff review before applying
- Built-in version control — branch, rewind, and diff any plan state
- Supports Anthropic, OpenAI, Google, xAI, OpenRouter, Ollama, DeepSeek, and more
- Tree-sitter powered project maps for 30+ languages (up to 2M tokens of context)
- Automated debugging for terminal commands and browser applications
- BYO API keys — LLM credentials stay on developer machines, never stored server-side

The template deploys a Go-based API server with an embedded Python LiteLLM proxy subprocess, backed by PostgreSQL for metadata and a persistent volume for plan file storage.

## Why Deploy Plandex on Railway

- One-click deploy with PostgreSQL and persistent storage pre-configured
- No Docker or server management — Railway handles scaling and TLS
- Keep your source code and API keys off third-party cloud services
- Team collaboration on AI plans with a shared self-hosted server
- Plandex Cloud is winding down — self-hosting is the primary path forward

## Common Use Cases for Self-Hosted Plandex

- **Large-scale refactoring** — plan and execute changes spanning dozens of files across big codebases with AI assistance
- **Full-feature development** — build entire features or applications from scratch with AI planning, implementation, and automated debugging
- **Codebase exploration** — project-aware chat mode for asking questions and understanding unfamiliar codebases
- **Team AI coding server** — shared self-hosted instance where multiple developers collaborate on AI-assisted plans

## Dependencies for Plandex on Railway

- **plandex-server** — `plandexai/plandex-server:latest` (Go API server + embedded LiteLLM Python proxy)
- **PostgreSQL** — Railway-managed database for metadata, migrations, and user data

### Environment Variables Reference for Plandex

| Variable | Description | Default |
|---|---|---|
| `DATABASE_URL` | PostgreSQL connection string | Required |
| `GOENV` | Server mode (`development` or `production`) | `development` |
| `LOCAL_MODE` | Skip email verification when `1` | `1` |
| `PLANDEX_BASE_DIR` | Plan file storage directory | `/plandex-server` |
| `PORT` | HTTP server listening port | `8099` |
| `API_HOST` | Public URL (required in production mode) | — |

### Deployment Dependencies

- **Runtime:** Go 1.23+ server, Python 3.x venv (LiteLLM)
- **Docker Hub:** [plandexai/plandex-server](https://hub.docker.com/r/plandexai/plandex-server)
- **GitHub:** [plandex-ai/plandex](https://github.com/plandex-ai/plandex)
- **Docs:** [docs.plandex.ai](https://docs.plandex.ai)

## Hardware Requirements for Self-Hosting Plandex

| Resource | Minimum | Recommended |
|---|---|---|
| CPU | 1 vCPU | 2 vCPU |
| RAM | 512 MB | 1 GB |
| Storage | 1 GB (server + plans) | 5 GB+ |
| Database | PostgreSQL 14+ | PostgreSQL 16 |

The Go server is lightweight. Resource usage scales with the number of active plans and users. LLM inference is offloaded to external providers — the server only routes API calls.

## Self-Hosting Plandex with Docker

Clone the repository and run with Docker Compose:

```
git clone https://github.com/plandex-ai/plandex.git
cd plandex/app
docker-compose up -d
```

Or run the server image directly:

```
docker run -d \
  -p 8099:8099 \
  -e DATABASE_URL="postgres://user:pass@db:5432/plandex?sslmode=disable" \
  -e GOENV=development \
  -e LOCAL_MODE=1 \
  -e PLANDEX_BASE_DIR=/plandex-server \
  -v plandex-data:/plandex-server \
  plandexai/plandex-server:latest
```

For production mode with email verification, set `GOENV=production` and configure SMTP variables (`SMTP_HOST`, `SMTP_PORT`, `SMTP_USER`, `SMTP_PASSWORD`).

## How Much Does Plandex Cost to Self-Host?

Plandex is fully open-source under the MIT License. There are no license fees, usage limits, or paid tiers for the self-hosted server. Plandex Cloud (the hosted option) is winding down and no longer accepts new users — self-hosting is the recommended path.

Your only costs are Railway infrastructure (typically $5–10/month for light usage) and LLM API fees from your chosen provider (OpenAI, Anthropic, etc.). Users bring their own API keys — the server never stores them.

## Plandex vs Aider vs Cursor

| Feature | Plandex | Aider | Cursor |
|---|---|---|---|
| Interface | Terminal CLI | Terminal CLI | GUI IDE |
| Multi-file changes | Yes (50+ files) | Yes | Yes |
| Plan versioning | Built-in branches/rollback | Git-based | No |
| Self-hosted server | Yes | No (local only) | No |
| Open source | MIT | Apache 2.0 | Proprietary |
| LLM providers | 11+ via LiteLLM | OpenAI, Anthropic, others | Built-in |
| Team collaboration | Yes (shared server) | No | Limited |

Plandex stands out with its server-based architecture enabling team collaboration and its built-in plan versioning system for reviewing and rolling back AI changes.

## FAQ

**What is Plandex and why self-host it?**
Plandex is an open-source AI coding agent that plans and executes multi-step development tasks across large codebases. Self-hosting keeps your source code, conversation history, and API routing on your own infrastructure instead of third-party servers.

**What does this Railway template deploy for Plandex?**
The template deploys a Plandex API server (Go + embedded LiteLLM Python proxy) with a Railway-managed PostgreSQL database and a persistent volume for plan file storage. The server listens on port 8099.

**Why does the Plandex Railway template include PostgreSQL?**
PostgreSQL stores user accounts, organization data, plan metadata, and handles database migrations. It is required — the server will not start without a valid database connection.

**How do I connect the Plandex CLI to my self-hosted Railway server?**
Install the CLI with `curl -sL https://plandex.ai/install.sh | bash`, then set `export PLANDEX_API_URL=https://your-domain.up.railway.app` and run `plandex sign-in` to create an account or log in.

**How do I enable production mode with email authentication on self-hosted Plandex?**
Change `GOENV` to `production`, remove `LOCAL_MODE`, set `API_HOST` to your Railway public URL, and configure SMTP variables (`SMTP_HOST`, `SMTP_PORT`, `SMTP_USER`, `SMTP_PASSWORD`). Users will receive PIN codes via email to verify their accounts.

**Can I use local LLMs like Ollama with self-hosted Plandex on Railway?**
The server supports Ollama via the `OLLAMA_BASE_URL` environment variable, but the Ollama instance must be network-accessible from Railway. You would need to host Ollama separately with a public endpoint or use Railway's private networking if both services are in the same project.


## Similar templates

- [Chat Chat](https://railway.com/deploy/-WWW5r) — Chat Chat, your own unified chat and search to AI platform.
- [stella](https://railway.com/deploy/stella) — Self-host stella with web, API, Postgres, Redis, and object storage.
- [Hermes Agent | OpenClaw Alternative with Dashboard](https://railway.com/deploy/hermes-agent-or-openclaw-alternative-wit) — Self-Hosted Hermes AI Agent for Telegram, Discord & Slack

Open this page in a browser: https://railway.com/deploy/plandex