Give your AI a small database of your own context on Railway

Ops & admin automation · Personal AI context · checked 2026-10-18

Run Postgres and n8n on Railway, let a compatible AI client propose one table at a time through a Postgres MCP server, and use n8n to keep it filled from the apps you already use. You approve every schema before it is created.

Independent review. IonBin has no partnership with any tool on this page and was not paid for this page. Our evaluation is our own, based on public documentation and our research. Product names are trademarks of their owners and are used for identification only. How disclosure works.

Goal

Let your assistant answer from your own tasks, decisions and notes instead of guessing.

Connects through: Built-in integration · MCP · API

Steps

  1. You

    List what you wish your assistant could see, then pick one table to start (tasks or decisions)

    Human stepConfirmed
  2. Railway

    Adds a Postgres database from + New or Cmd/Ctrl+K and exposes DATABASE_URL

    Built-in integrationConfirmedSource ↗
  3. Railway

    Deploys n8n from a one-click template

    Built-in integrationConfirmedSource ↗
  4. Your AI platform

    Uses a compatible local/stdio or SSE client to connect through Postgres MCP Pro and propose the table design

    Unrestricted mode allows reads and writes; restricted mode limits the server to read-only transactions.

    MCPConfirmedSource ↗
  5. You

    Approve the schema before anything is created

    Human stepConfirmed
  6. n8n

    Syncs one source into the table on a schedule with the Postgres node

    APIConfirmedSource ↗

Confirmed means the vendor documents it (linked). It doesn't mean IonBin has tested it. AI-platform and human steps need no source.

Human approval

You approve every schema change; agents that only read get a read-only connection.

Watch out for

A database with write access is a real system. Enable backups and verify a restore, give read-only agents a restricted connection, and don't store anything you couldn't afford to leak.

Starter prompt

Using the Postgres MCP connection, propose a schema for two tables: decisions (what, why, date, status) and tasks (title, due, status, linked decision). Explain each column and index. Do not create anything until I say go.

Written for this workflow. Paste it into your AI platform and adjust it to your situation.

Tools in this workflow

ToolRoleOfficial site
RailwayBuilt-in integration railway.com ↗
n8nAPI n8n.io ↗
ClaudeMCP claude.ai ↗

Which AI platforms can run it

PlatformStatusWhy
Claude (incl. Claude Code) To verify Your AI platform: MCP pairing needs route-specific evidence; a named plugin or CLI does not establish it; n8n: verify the API caller, network access and credentials; platform code execution alone does not establish this connection
OpenAI Codex To verify Your AI platform: MCP pairing needs route-specific evidence; a named plugin or CLI does not establish it; n8n: verify the API caller, network access and credentials; platform code execution alone does not establish this connection
Google Antigravity To verify Your AI platform: MCP pairing needs route-specific evidence; a named plugin or CLI does not establish it; n8n: verify the API caller, network access and credentials; platform code execution alone does not establish this connection
ChatGPT To verify Your AI platform: MCP pairing needs route-specific evidence; a named plugin or CLI does not establish it; n8n: verify the API caller, network access and credentials; platform code execution alone does not establish this connection
Gemini To verify Your AI platform: MCP pairing needs route-specific evidence; a named plugin or CLI does not establish it; n8n: verify the API caller, network access and credentials; platform code execution alone does not establish this connection
Perplexity To verify Your AI platform: MCP pairing needs route-specific evidence; a named plugin or CLI does not establish it; n8n: verify the API caller, network access and credentials; platform code execution alone does not establish this connection

Readiness considers the connection evidence and unresolved steps. Confirmed: documented access and the proposed steps meet the recorded requirements. MCP and API pairings require an explicit reviewed platform note; a named plugin or CLI alone does not confirm MCP. Inferred: an explicit page note describes a plausible pairing that still needs testing. To verify: a connection or step is unresolved. An API and code execution alone do not establish compatible network access, credentials or permissions. None of this is IonBin testing.

What we checked (2026-10-18)

  • Railway. You add Postgres from the Cmd/Ctrl+K menu or + New on the project canvas, and other services in the project use DATABASE_URL. The database is private by default. Adding public access creates a TCP proxy and a DATABASE_PUBLIC_URL, and egress through that proxy is billed.
  • n8n on Railway. Railway’s guide covers a one-click n8n template, or a manual setup from the official Docker image with a volume. n8n’s Postgres node can insert, update and query rows.
  • MCP servers. Postgres MCP Pro (Crystal DBA) has an unrestricted read/write mode and a restricted, read-only mode. n8n-MCP helps an agent work with n8n node documentation and design workflows. These are third-party projects, not official PostgreSQL or n8n products. A star count is not evidence of reliability.
  • Connecting your assistant. Claude’s custom connectors use remote MCP servers, available on Claude, Cowork and Claude Desktop (Free accounts get one). Postgres MCP Pro’s documented setup uses local stdio or SSE. That does not establish a hosted Claude connector or support in clients that require Streamable HTTP. Keep the database private and use a compatible local client or an authenticated server deployment with private database access.
  • Alternatives. Supabase offers an official MCP server with project scoping and read-only SQL mode. Restrict feature groups too; read-only SQL is not a blanket promise that every other exposed tool has no side effects.

Set it up

  1. Start with one table, and add more once you use it.
  2. Approve each schema in writing before the agent creates it.
  3. Add one sync at a time, and watch it for a week.

Credit: the idea comes from a Substack podcast episode about a personal database by creator Angus Sewell. IonBin reviewed the cited technical documentation. The assembled workflow has not been tested end to end; the inspiration link is credit, not evidence of performance.

Verification limits and implementation checks

The page is documentation-reviewed; the complete workflow has not been run by IonBin.

  • Test the selected client transport and authentication before exposing a remote server.
  • Inspiration article could not be retrieved during this audit; it is not used to establish technical claims.

Sources

Related workflows

No affiliate links on this page. No paid placements in workflows. How IonBin makes money.