shipwithjev

Catalog / Tools & apps

0148GitHub

maza

A local MCP gateway with Jev tool discovery and a dashboard.

prasanth263/mazaREADME ↗
# Maza

Maza is a local MCP gateway. Connect your MCP servers once, then point Claude Code or another MCP client at Maza. Agents see two tools: `find_tool` discovers a capability through Jev Choice, and `execute_tool` runs it after the agent supplies arguments from the original schema.

The gateway manages connections, credentials, discovery, per-tool access and execution. It runs as a native Node.js process; Docker is not required. Jev is the gateway's only model integration. Planning and argument generation remain with the calling agent.

## Requirements

- Node.js 22.13 or later and npm.
- An available OS credential store supported by `@napi-rs/keyring`. macOS Keychain has been tested. Linux/Windows credential-store integration has not been verified.
- A Jev API key from TypeSafe AI.
- Claude Code installed if you want automatic Claude registration.
- Each downstream MCP server's runtime and credentials, when required.

## Install and start locally

```sh
git clone https://github.com/prasanth263/maza.git
cd maza
npm ci
npm link
maza init
maza secret set jev
maza start
```

`npm link` makes the `maza` command available locally on your machine. If you prefer not to link it, replace `maza` with `node /absolute/path/to/maza/src/cli.js` in the commands below, or use `npm run maza -- <command>` from this repository.

`maza secret set jev` prompts for the key with hidden input. Paste the key and press Enter. The value is stored in OS credentials, not in the repository, shell arguments or configuration file. To replace the key later, run the same command; no daemon restart is needed solely for a key change. Missing or placeholder keys do not trigger a mock routing fallback.

By default, state is stored in `~/.maza`. To use another directory, set `MAZA_HOME` before running co

Also filed under Tools & apps

  1. 0616

    Jevstiller: distill Jev calls into a local model

    A self-hosted proxy that trains a local model on Jev's answers. It served 72% of Banking77 locally at 15 ms, with a 2% audit and a bound on disagreement.

    tomerglick57 · Tools & apps · 15 ms local

  2. 0611

    Scruple: code checks for what linters miss

    Semantic-aware linting with System One models: rules you can state in a sentence, checked against your code by Jev.

    @nalexpeae · Tools & apps

  3. 0606

    Algolia-style Postgres search with a Jev reranker

    Typo tolerance, typeahead, barcodes and facet counts in plain Postgres SQL, plus an optional Jev step that asks one question per top-10 result.

    @alex_forman1 · Tools & apps

  4. 0604

    Virtual Closet: outfits from closet videos

    built a virtual closet & outfit generator from raw closet videos. classification + semantic search thanks to jev 🥰. clueless closet dreams finally coming to life

    @maitreemervana · Tools & apps