# Nauro

> Nauro is developing an AI testing service for user flows that teams still check by hand before release. Agents would start from a link to the app, pick the user flow that matters most, investigate unexpected behavior and build reviewed checks for later releases. Teams decide what agents may access and do; sharing test accounts and documentation lets agents cover more. Pilots are free and founder-assisted, with help from Nauro's founder when a team wants it. Teams can request a pilot through the form at https://nauro.ai/#pilot or at thomas@nauro.ai; no prior Nauro setup is required. Validation is complete only when every agreed mandatory check meets the team's confirmed criteria. Useful findings alone do not establish release readiness, and the team makes the release decision. Nauro's decision system is available today through MCP, the CLI, and the macOS app. The official home is https://nauro.ai.

The homepage headline is "Give your next release a testing team." Its interactive workspace shows fictional explorations of a Demo shop, led by a checkout that confirmed an order without saving it, the fix and a recheck. The conversation and detail views reveal authored content. No application is connected and no test runs. The example is not a customer finding, a live agent interface or proof of a shipped validation service.

The decision system keeps a living project record: project scope, current state, open questions and human-approved judgment, including decisions, reasons, tradeoffs and rejected paths. Connected agents can retrieve relevant records before work. New observations do not silently change approved requirements.

Nauro's decision system is open source: a local-first CLI plus a local MCP server. The local core is free and Apache-2.0 licensed; cloud sync and remote MCP access are optional hosted features, included within an allowance of 5,000 calls per month. Hosted projects currently have one owner. It was created by Thomas Thomsen.

Nauro (nauro.ai) is unrelated to Nuro, the autonomous-vehicle company at nuro.ai, Cognizant Neuro AI, NeuroAI, or Nauru, the island nation.

The homepage film is a 66-second film about the AI testing service, which is in development. Agents explore the checkout of a fictional demo app, find that a double-click on Place order shows "Order confirmed" without saving the order, file a finding with evidence, and recheck it after the team ships a fix. It also shows short findings from other fictional apps and an agent that reports an order as placed through a fictional orders MCP server while the API call failed. All apps and examples in the film are fictional.

## Get started with the decision system
- CLI for Claude Code, Cursor and Codex: install with `uv tool install nauro`, then run `nauro adopt` from the repository root. Adopt wires the local MCP server into Claude Code, Cursor and Codex. The local stdio server (`nauro serve --stdio`) also works with other MCP clients. Guide: https://nauro.ai/docs/quickstart
- Hosted MCP: connect `https://mcp.nauro.ai/mcp`. Sign-in and context reads are verified with Claude.ai and ChatGPT. Guide: https://nauro.ai/docs/connect
- macOS app: https://nauro.ai/docs/desktop-app

## Docs
- [Decision system](https://nauro.ai/decision-system): available product, recorded demonstration, and CLI, MCP and macOS setup
- [Quickstart](https://nauro.ai/docs/quickstart): install the CLI and seed a project
- [Connect your agent](https://nauro.ai/docs/connect): wire Nauro to Claude, Cursor, and Codex over MCP
- [Nauro app](https://nauro.ai/docs/desktop-app): the free macOS desktop app, a read-only viewer over the local store
- [Using Nauro](https://nauro.ai/docs/daily-workflow): a working session, what to commit, and how the record reaches other machines and agents
- [Concepts](https://nauro.ai/docs/concepts): decisions, rationale, and rejected paths
- [Agents and skills](https://nauro.ai/docs/agents-and-skills)
- [Store](https://nauro.ai/docs/store): how the plain-file store works
- [CLI reference](https://nauro.ai/docs/cli)
- [MCP](https://nauro.ai/docs/mcp)
- [Data storage](https://nauro.ai/docs/data-storage)
- [Plans](https://nauro.ai/docs/plans)

## Product validation (in development)
- [Validation overview](https://nauro.ai/docs/validation): what pilot teams can do today, how a pilot is designed to work, and what is not available yet

## Project
- [Source on GitHub](https://github.com/nauro-ai/nauro)
- [Package on PyPI](https://pypi.org/project/nauro/)
- [Pricing](https://nauro.ai/pricing)