Railway Operator Public

railway-operator

[](https://opensource.org/licenses/MIT) [](https://claude.com/claude-code) [](https://www.anthropic.com/claude)

A [Claude Code](https://claude.com/claude-code) plugin that dispatches an autonomous agent, running on **the session model** (always the strongest available Claude — no hardcoded model dependency), to execute any Railway infrastructure task end-to-end. The agent follows a strict escalation ladder: **CLI first**, **GraphQL API** when the CLI lacks a mutation, **Playwright** against the Railway dashboard for UI-only actions. Every mutation is verified, every known gotcha is applied automatically, and authentication remains under the operator's control.

Current version: **0.1.1**.

What it does

When you invoke the `railway-op` skill (or ask Claude to do something on Railway), the plugin:

1. Context gathering — captures Railway auth state, project link, git state, project type hints
2. Agent dispatch — a fresh-context agent (running on the session model) with the full Railway operating knowledge baked in
3. Execution — the agent plans, executes, and verifies each step using the escalation ladder, ruling out operator error before escalating a rung
4. Structured report — task, actions taken, verification results, gotchas avoided, follow-up items

Installation

# 1. Add this repo as a marketplace
claude plugin marketplace add https://github.com/TheMizeGuy/railway-operator-public.git

# 2. Install the plugin
claude plugin install railway-operator@railway-operator-public

# 3. Restart Claude Code for the plugin to load

After restart, verify with `claude plugin list` and look for `railway-operator@railway-operator-public`.

Prerequisites

• Railway CLI ≥ 4.36.0, installed with authentic