Reference

Scripts

Script	Purpose
pnpm run dev	Run Vite and the API server for local web development.
pnpm run dev:api	Run only the Express API server.
pnpm run start	Serve the built app/API.
pnpm run build	Typecheck and build the Vite app plus CLI bundle.
pnpm cli:build	Build the Node CLI to dist/cli.js.
pnpm cli:dev	Run the TypeScript CLI entry with tsx.
pnpm run test	Run Vitest tests.
pnpm run lint	Run Oxlint.
pnpm run format	Check formatting with oxfmt.
pnpm desktop:dev	Run the Tauri desktop app in development.
pnpm desktop:check	Typecheck and run Rust cargo check.
pnpm desktop:build	Build the Tauri desktop app.
pnpm eval:smoke	Run the Promptfoo smoke eval.
pnpm eval:iterations	Run focused iteration evals.
pnpm eval:iterations:matrix	Run the full iteration matrix.
pnpm docs:dev	Run this documentation site locally.
pnpm docs:build	Build this documentation site.
pnpm docs:preview	Preview the built documentation site.

Provider Settings

See Provider Setup for LM Studio startup steps, UI endpoint configuration, OpenAI-compatible endpoint requirements, and remote-provider limitations.

type ModelProviderSettings = {
  baseUrl: string;
  model?: string;
  reasoningEffort?: 'none' | 'minimal' | 'low' | 'medium' | 'high' | 'xhigh';
};

Default:

{
  baseUrl: 'http://localhost:1234/v1',
  model: 'gemma-4',
  reasoningEffort: 'none',
}

Pipeline Options

type PipelineOptions = {
  includeDebug?: boolean;
  styleThreshold?: number;
  finalSmoothing?: boolean;
  maxRewriteIterations?: number;
  runMeaningCheck?: boolean;
};

GET /api/health

Checks whether the default provider, or a supplied baseUrl, can list models.

curl http://127.0.0.1:5174/api/health

curl 'http://127.0.0.1:5174/api/health?baseUrl=http://localhost:11434/v1'

Response fields:

{
  gemma4Found: boolean;
  lmStudioReachable: boolean;
  model: string;
  ok: boolean;
  status: string;
  error?: string;
}

GET /api/models

Lists models from the default provider or a supplied baseUrl.

curl 'http://127.0.0.1:5174/api/models?baseUrl=http://localhost:1234/v1'

POST /api/rewrite

Runs the rewrite pipeline.

curl -X POST http://127.0.0.1:5174/api/rewrite \
  -H 'Content-Type: application/json' \
  --data '{
    "document": "Our platform leverages advanced AI to improve workflows.",
    "provider": {
      "baseUrl": "http://localhost:1234/v1",
      "model": "google/gemma-4-12b-qat",
      "reasoningEffort": "none"
    },
    "options": {
      "includeDebug": true,
      "maxRewriteIterations": 2,
      "runMeaningCheck": true
    }
  }'

POST /api/eval/rewrite

Runs the eval rewrite endpoint against fixture profiles.

{
  "source": "This paragraph needs a clearer product-note style.",
  "styleProfileId": "direct-technical",
  "providerId": "lmstudio",
  "model": "google/gemma-4-12b-qat",
  "options": {
    "maxRewriteIterations": 2,
    "reasoningEffort": "none",
    "runMeaningCheck": true,
    "runFinalSmoothing": false
  }
}

Fixture profiles live under evals/fixtures/profiles/.

Eval Environment Variables

Variable	Purpose
STYLEMAKAR_API_BASE_URL	StyleMakar API server URL. Defaults to http://127.0.0.1:5174.
STYLEMAKAR_EVAL_BASE_URL	OpenAI-compatible provider base URL for eval scripts.
STYLEMAKAR_EVAL_PROVIDER_ID	Provider id or URL sent to /api/eval/rewrite; takes precedence over STYLEMAKAR_EVAL_BASE_URL.
STYLEMAKAR_EVAL_MODEL	Model id sent to the provider.
STYLEMAKAR_EVAL_REASONING_EFFORT	Reasoning effort passed through to compatible providers. Defaults to none.
STYLEMAKAR_EVAL_CASE_FILTER	Filter used by the matrix eval script.