---
title: "Deploy Mastra workflow evaluation demo"
description: "Deterministic Mastra memory and human approval workflow demo."
category: "AI/ML"
url: https://railway.com/deploy/mastra-workflow-evaluation-demo
---

# Deploy Mastra workflow evaluation demo

Deterministic Mastra memory and human approval workflow demo.

**[Deploy Mastra workflow evaluation demo on Railway](https://railway.com/template/mastra-workflow-evaluation-demo)**

Machine-readable deploy manifest (JSON, validated by TemplateCI): https://railway.com/deploy/mastra-workflow-evaluation-demo/manifest.json

- **Creator:** Anton Orel's Projects
- **Category:** AI/ML

## Template content

### Mastra API

- **Source:** tech-progress/mastra-agent-postgres
- **Start command:** `node .mastra/output/index.mjs`
- **Health check:** /readyz
- **Public domain:** Yes

### Postgres

- **Image:** postgres:17.11-bookworm@sha256:639ab7ceb90e13123085b741fb31ef493fba25463002f6da665352e7b534b652

## Documentation

# Deploy and Host Mastra workflow evaluation demo

Deterministic Mastra memory and human approval workflow demo.

## About Hosting Mastra workflow evaluation demo

This single-operator evaluation recipe combines a typed [Mastra](https://mastra.ai/) API with private [PostgreSQL](https://www.postgresql.org/) storage. It persists conversations and workflow snapshots, and demonstrates explicit human approval before a SQL-idempotent ledger effect. It does not ship public Studio or claim automatic recovery of arbitrary in-flight work.

## Why Deploy Mastra workflow evaluation demo on Railway?

Separate the public authenticated API from private persistent storage. Railway supplies service references and a database volume; the default deterministic fixture makes no model calls and is not an LLM. Optional hosted integration is UNVERIFIED and requires an operator-owned authorized key, with provider charges. A deliberately pinned Apache-only Mastra release and artifact scans bound the distributed package surface.

## Common Use Cases

- Evaluate native Mastra conversation persistence without a model bill.
- Evaluate human approval and explicit restart/resume workflows.
- Study an idempotent database tool before integrating external systems.

## Dependencies for Mastra workflow evaluation demo

- An accessible GitHub source authorized for Railway, with explicit repository, branch and root directory.
- Default `offline-fixture` mode needs no provider key and is not an LLM. Optional unverified OpenAI mode requires a real operator-owned key; calls incur charges.
- A generated bearer token, private PostgreSQL credentials and a persistent database volume.

### Deployment Dependencies

Mastra core 1.0.4 with compatible pinned tooling/adapters, Node 24.20.0 and PostgreSQL 17.11; dependency resolution and image inputs are locked. There is no bundled enterprise auth/editor or Studio UI. This is not an HA, production or multi-user deployment. Review the README, support boundary and artifact review before adoption.


## 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/mastra-workflow-evaluation-demo