DataNyx¶
Let an AI agent work with your data without ever seeing it.
DataNyx turns files and databases into value-free schemas an agent can write code against, and replaces names and identifiers in free text with placeholders. Start with Getting started, or pick the job you have:
| I want to… | Read | Command |
|---|---|---|
| Show an agent the shape of a file or database | Schemas | datanyx schema |
| Let it try code on realistic fake rows, clean or messy | Synthetic data | datanyx synth |
| Let it fix a script that fails on the real data, or check what it wrote | Debugging | datanyx run |
| Catch wrong column names before anything runs | Checking code | datanyx check |
| Find personal data or secrets in a folder before sharing it | Scanning | datanyx scan |
| Give it text with the names and ids removed | Text anonymization | datanyx anonymize |
| Share a copy of a table with consistent fake values | Pseudonymizing tables | datanyx pseudonymize |
| Measure and improve what gets detected | Detection quality | datanyx eval |
| Notice that a data feed changed shape | Schema diff | datanyx diff |
| Stop an agent from reading the data directly, or serve it over MCP | Using it with an agent | datanyx protect, datanyx mcp |
| Run it in a container, a pre-commit hook, CI or a shared server | Installing and deploying | datanyx mcp --http |
| Share one policy across a team | Team settings | datanyx.toml |
Where things are¶
- Features: what it does, in one page.
- Supported sources: file formats and databases, and the extra to install for each.
- Privacy model: what is guaranteed, what is only best effort, and the known limits.
- Command reference: every command and option, generated from the code.
- Python API: using DataNyx from your own code.
- Contributing: the architecture, how to add a source, and how to run the tests.