Privacy

noomoo is a command-line tool and a set of checks that run on your machine, in your repository. This document describes what it does with your data. It is accurate for the public package, which contains no code that sends telemetry.

What noomoo does not do

• It does not send usage data, analytics or crash reports anywhere.
• It does not contact any server operated by the noomoo authors.
• It does not upload your source code or documents to any service of its own.

What it writes on your machine

• Files inside your repository: documents, templates, hooks and checks that the commands you run create (docs/, .agent/, config/, .githooks/ and similar).
• A local run log, docs/validation/events.jsonl, to which each command appends the sub-command, exit code, duration, noomoo version and npm script name. It stays in your repository and is never sent anywhere.
• Two files in a user-level state directory (~/.config/noomoo, or the directory named by NOOMOO_CONFIG_DIR): trusted-repos.json, the list of repositories you agreed noomoo may run project code in, and identity-cache.json, a ten-minute cache holding your GitHub login, a short fingerprint of the token used to look it up and the time of the lookup. Neither file is sent anywhere.

Which commands talk to other services

noomoo opens no network connection of its own except the GitHub API calls listed here. Everything else is local.

• GitHub API (api.github.com). The commands and CI checks that look up a pull request, a merge status or the identity of a reviewer call the GitHub API with a token you or your CI provide (GITHUB_TOKEN, or the token of your gh login). noomoo does not store the token and does not forward it anywhere else.
• Tools you already have installed. Some commands run git (including git fetch), gh or claude on your machine: for example phase-check asks gh for the state of the current pull request, and identity asks gh for your token. These tools talk to their own services under your own accounts and are governed by their own terms and privacy policies.
• Model-based review. The commands task-ready, plan-ready (with --judge or --strict) and estimate start the Claude CLI installed on your machine and pass it the text of the document under review, with the task context and the rubric. They run only when you run them. The Claude CLI runs as an agent in the directory of your repository, so it can read other files of the repository with its own read tools; noomoo starts it with only the read tools (Read, Grep, Glob), without MCP servers, with hooks switched off and without the settings files of the repository, so the CLI has no tool to write files, run commands or fetch from the network. Anything it reads, including secrets such as a .env file in the repository, goes to the provider of the Claude CLI with the rest of the conversation. Whether it may read outside the repository depends on the permission rules of your own Claude CLI settings. It also inherits your environment, including any token in it. What happens to the text afterwards is governed by your agreement with the provider of the Claude CLI.

Repositories you do not trust

Until you run npx noomoo init --yes (or npx noomoo setup --yes) in a repository, noomoo does not load check modules defined by that repository's config/pipeline.json and npx noomoo doctor does not call gh there. Installing noomoo is not an agreement to run a repository's code. The other commands above run when you invoke them, in any repository.

The noomoo.dev website

This section is about the website, not the package. The package still sends no usage data or analytics.

• The pages of noomoo.dev count visits with Umami, an open-source analytics tool that runs on our own server. No third-party analytics service is involved, and nothing is shared with or sold to anyone.
• It sets no cookies and does not track you across other websites. Do Not Track is not honoured: every visit is counted the same way.
• For each page view it records the page address (including campaign parameters such as utm_source), the page title, the address of the referring page, the browser, operating system, device type, screen size, language, and the country, region and city derived from your IP address. It also records page load metrics: how fast the page rendered and responded.
• It records a few actions on the pages: clicks on install and GitHub links and on links to other websites (with the link text, the page and, for other websites, the target address), clicks on our email address, and the text of a command copied from a code block.
• The IP address itself is not stored. It is used only to derive the location and an anonymous visitor identifier, a hash of the IP address, the browser and a salt that changes every month.
• Like any web server, ours keeps standard access logs (time, IP address, requested address, referrer and browser) for operating and securing the site.

Contact

Questions about privacy: hello@noomoo.dev.