railwayapp/railway-skills
14 skills17.4k installs3.8k starsGitHub
Install
npx skills add https://github.com/railwayapp/railway-skillsSkills in this repo
1Use RailwayUse Railway is an agent skill for solo builders running MongoDB on Railway who need production-grade analysis without guessing from dashboards alone. It walks through what to collect over SSH with mongosh—server status, per-database and per-collection stats, active operations, profiler-backed slow queries, replication oplog health, and top read/write collections—and aligns those signals with Railway API infrastructure metrics for the same deployment. The skill encodes MongoDB-specific performance patterns such as WiredTiger cache pressure, checkpoint lag, ticket starvation, and scan-heavy query inefficiency, with fixes framed in terms indie operators can act on (more RAM, index work, profiling). It references analyze-db.md for consistent output structure and collection-status handling so agent reports stay comparable run to run. Reach for it when latency spikes, connections climb, or storage growth outpaces expectations on a Railway Mongo workload—not when you are only provisioning a new database from scratch.4kinstalls2ServiceRailway service is an agent skill for operators and solo builders who already use Railway and need precise control over individual services without guessing CLI coverage gaps. It covers reading service status and deployment health, updating presentation and linkage, and creating services through the GraphQL API when the CLI cannot. The documentation steers the common path—scaffolding local code with the new skill and pointing GitHub repositories through new plus environment—while this skill stays focused on inspection, property updates, and Docker-based service creation. Bash is constrained to railway:* commands, so agents follow a predictable, platform-native workflow. You get explicit mutations, context extraction patterns, and cross-skill handoffs instead of ad-hoc dashboard hopping.1.6kinstalls3Central StationCentral Station is a Railway agent skill that wires your coding agent into Railway’s community support platform—Central Station—so you can search threads, skim discussions, and pull markdown exports before duplicating questions or misreading product docs. Solo builders on Railway use it when they ask to search Central Station, find what people are asking about deployments or databases, or review recent support patterns. The skill documents concrete HTTP endpoints for GraphQL queries, per-thread markdown, and bulk LLM-oriented exports, with Bash curl and jq as the execution path. It deliberately defers official product documentation to the railway-docs skill and live project health to status, keeping this skill in the research lane. Multi-phase value shows up when you discover prior art in Idea, unblock Build integrations, or triage Operate support themes. Beginner-friendly if you are comfortable running curl from the agent.1.5kinstalls4StatusRailway Status is an integration skill for solo builders using Railway who need a fast, JSON-backed picture of the linked project before touching deploys or services. It tells the agent when to run `railway status --json`, how to confirm the CLI is installed and authenticated, and how to recover from common setup gaps with copy-paste install and login steps. The skill deliberately scopes itself away from environment-variable and builder-configuration questions, pointing those to the companion `environment` skill so agents do not mix “what’s deployed” with “how is it configured.” Use it at the start of a Railway session, when someone asks if the app is running or what’s live, or immediately before deploy, variable, or domain operations so downstream commands hit the right project context. For Prism’s journey, it bridges Ship prep and Operate visibility without replacing full observability stacks.1.5kinstalls5Railway DocsRailway Docs is an agent skill for solo builders who host APIs, web apps, or workers on Railway and cannot rely on model memory for fast-moving platform details. It instructs the agent to pull current documentation from Railway’s LLM-oriented endpoints—full llms-docs, llms.txt index, templates, changelog, and blog—and to convert standard docs URLs to `.md` variants for accurate retrieval. Use it when someone asks how Railway projects, deployments, volumes, or pricing work, shares a docs.railway.com link, or needs verification before changing production config. The skill is deliberately narrow: it is a research-and-answer layer, not a deploy script. It fits Operate when tuning infra, but equally supports Build and Ship when wiring services and env vars. Always prefer fetched docs over hallucinated platform behavior.1.1kinstalls6DeploymentThe deployment skill packages Railway’s CLI for solo builders who already host on Railway and need fast operator actions from Claude Code, Cursor, or Codex. It covers the deployment lifecycle—list history, check status, redeploy, restart, stop, or remove the active deployment—plus log visibility for debugging failed builds or crashing releases. The description explicitly steers you away from service deletion: tearing down a deploy is not the same as deleting a service, which belongs to the environment skill with `isDeleted: true`. Triggers match plain language like “redeploy,” “show logs,” and “why did deploy fail,” so agents invoke it during incidents rather than during initial project scaffolding. Expect shell and network access only; you need the Railway CLI linked to the right project. It is a focused integration skill, not a full CI/CD design guide.1kinstalls7DeployDeploy is a Railway-focused agent skill that standardizes how you ship code with the Railway CLI. It tells the agent to use `railway up` when you say deploy, ship, push, or railway up, always passing a descriptive `-m` message that summarizes what is going out. Solo builders benefit because deploy steps stop being ambiguous chat suggestions and become repeatable bash commands with two clear modes: detach for routine releases and CI mode when you need to watch logs or fix build breaks. The skill also signals when to pick a service, when to escalate to the new skill for greenfield setup, and when environment configuration matters for Docker images. It is narrow by design—deployment only—which keeps it reliable in Claude Code sessions that already allow `Bash(railway:*)`.959installs8DatabaseDatabase is a Railway-focused agent skill for solo and indie builders who deploy on Railway and need a managed datastore without hand-rolling Docker or connection secrets. It walks your agent through the correct order of operations: inspect the project for an existing database service by querying environment config for source.image, create an official Railway Postgres, Redis, MySQL, or MongoDB template only when none exists, wait for the deployment to become ready, and optionally connect your application by delegating variable wiring to the env skill. The skill is intentionally narrow compared to the broader templates skill, which covers Ghost, Strapi, n8n, and other non-database stacks. It fits Claude Code, Cursor, and Codex sessions where Bash access to railway:* is allowed, so the agent can run real CLI commands instead of guessing connection strings. Use it when you say add postgres, add redis, connect to database, or wire up the database alongside an existing server service.890installs9ProjectsProjects is a Railway CLI agent skill for solo builders who already deploy on Railway and need to orient across workspaces, pick the right linked project in a repo, or adjust platform settings from natural-language requests. It encodes when to list all projects, switch projects, rename, tune PR deploy behavior, or change public versus private visibility. The workflow favors subagent-friendly JSON commands, then returns simplified summaries rather than raw payloads. After a switch it points you at the companion status skill for deeper project detail. It fits operators and indie hackers juggling multiple apps on one account who want the agent to run `railway` commands safely under the allowed-tools policy instead of guessing CLI flags.867installs10EnvironmentRailway Environment is a narrow integration skill that teaches your agent to read and edit Railway environment configuration through the official CLI instead of guesswork in chat. Solo builders use it when they need to see what is actually deployed—which repo branch, which build command, which variables each service sees—or when they want to add, set, or delete variables, tweak replicas, or clone production into staging. The workflow centers on JSON dumps from railway environment config so you can present a structured picture of the stack, then apply changes the user requests with the same tool surface. It fits indie SaaS and API projects already on Railway where environment drift causes outage-grade surprises. Complexity stays beginner-friendly as long as the CLI is installed and linked to the right project; the skill does not replace Terraform but complements day-two ops for Railway-native hosting.841installs11Metricsmetrics is a Railway-focused agent skill that teaches solo builders how to pull structured resource measurements for services already running on Railway. After linking a project, you derive environment and service identifiers from railway status JSON, then query measurements such as CPU cores used versus limited, memory and network throughput in gigabytes, and ephemeral or backup disk consumption. The skill documents grouping dimensions so you can compare deployments, instances, or regions when isolating regressions. Pair it with the service skill when someone asks broadly whether a workload is healthy versus merely under-provisioned. It assumes the Railway CLI is installed and the workspace is linked; it does not replace application-level tracing or log correlation but gives fast infra-level signals for operate-phase triage.811installs12NewRailway new is an agent skill for solo builders who want Railway hosting without memorizing CLI sequences. It activates when you say deploy to Railway, create or link a project, add a backend or API service, scaffold from a template stack, or wire a GitHub repository into a new service. The skill treats initial setup and incremental service adds as one workflow: confirm the CLI is installed and authenticated, then drive project initialization, linking, and service creation with allowed Bash scoped to railway, which, command, npm, and npx. It is aimed at indie shipping loops where the agent already wrote the app and you need production placement fast. Database provisioning is intentionally out of scope here—you pair this skill with the Railway database skill when persistence is required. Expect shell execution against your machine and Railway account; review commands before approving in untrusted environments.801installs13DomainThe Railway domain skill tells your coding agent exactly how to expose a deployed Railway service on the internet. It covers generating the platform subdomain (one per service), attaching custom domains with JSON DNS guidance, and discovering what is already configured through environment config or GraphQL. Solo builders use it when they have a running deployment but no shareable URL, when they are wiring example.com to production, or when they need to audit or remove hostnames. Commands are Bash-scoped to railway:* so the agent does not improvise unrelated shell. Pair it with Railway deploy and environment skills so domain steps run only after a successful deployment.783installs14TemplatesRailway templates is an integration skill for solo builders shipping on Railway who want marketplace services in minutes instead of bespoke infra repos. It triggers when someone says add Ghost, add n8n, find a CMS template, or list what templates exist, while nudging pure database provisioning to the sibling database skill for clearer boundaries. The workflow starts with project context from railway status --json—project id and default environment id—then uses documented template codes or search for less common stacks. Categories span databases (when not delegated), CMS, object storage, workflow automation, and uptime monitoring, matching typical indie SaaS footprints. Because it runs through the Railway CLI and API scripts, the agent stays inside permitted railway:* bash tooling. Use during Build when composing your stack, and again in Validate when you need a quick prototype of Strapi or n8n before committing architecture. Outcome is a deployed template service wired into the same Railway workspace you already use for your app.729installs