Nauro

Decision system for AI coding agents. Surfaces what your project decided, and why, before changes.

Nauro is an MCP server designed specifically for AI coding agents working inside software projects. Its core function is to act as a decision system: it records what a project has decided and why, then makes that context available to an agent before it attempts to modify code. This addresses a common failure mode in AI-assisted development where an agent introduces changes that contradict prior architectural, design, or implementation decisions because it had no access to that reasoning.

The server communicates over stdio transport and is distributed as a Python package on PyPI under the identifier nauro at version 1.2.1, published by ai.nauro. The repository is hosted at github.com/Nauro-AI/nauro.

Because no tool list is exposed in the registry record, the exact set of operations the server provides — such as querying decisions, recording new ones, or tagging rationale to specific files or modules — cannot be confirmed from available facts. What is documented is the stated purpose: surfacing project decisions and their reasoning to a coding agent as a pre-change consultation step.

This server is appropriate when a project maintains a structured record of decisions and needs those records injected into an agent’s context automatically. It is not appropriate for projects that do not maintain decision records, for non-coding agent workflows, or for environments where stdio transport is not supported.

Use cases

  • Surface architectural decision records to a coding agent before it refactors a module
  • Prevent an agent from reintroducing a pattern the team explicitly decided to abandon
  • Provide rationale context to an agent generating new code in an established codebase
  • Consult recorded decisions when an agent is resolving a dependency or framework choice

When to use it

  • The project maintains structured records of past decisions and their rationale
  • An AI coding agent is being used to modify an existing codebase with documented constraints
  • The development environment supports stdio transport for MCP servers
  • The team wants decision context injected automatically before agent-driven changes

When not to use it

  • The project has no recorded decisions for the server to surface
  • The agent workflow is not related to code editing or software development
  • The runtime environment does not support stdio transport
  • A tool-level API is required and the available tool list needs to be verified before adoption