Getting Started

Requirements

• Node.js 18 or newer
• npm
• Git
• Relay or another OpenAI-compatible local server

Install From Source

git clone git@github.com:achuthanmukundan00/synax.git
cd synax
npm install
npm run build

During local development, run the built CLI through the package script:

npm run synax -- --help

After publishing or linking the package, the command name is:

synax

Create Project Config

Synax works with defaults, but a .synax.toml makes provider and verification behavior explicit:

cp .synax.toml.example .synax.toml

Edit the model name to match the model exposed by Relay.

Check The Setup

Run a quick local health check:

npm run synax -- doctor

Run the full provider check after Relay is running:

npm run synax -- doctor --full

Full doctor mode probes /models and sends a small /chat/completions request. Quick mode skips live provider calls.

First Session

npm run synax -- inspect
npm run synax -- ask --question "Summarize this repository in five bullets."
npm run synax -- chat

Inside chat:

synax> /settings
synax> /tools
synax> /budget
synax> /test-provider
synax> Explain how config loading works.
synax> /verify
synax> /exit

For one bounded edit-capable task:

npm run synax -- run --task "Fix the failing test"

synax run --plan plan.md is accepted by the CLI, but the plan execution engine is still a placeholder.