# Brandkit

Brandkit is a public API for extracting canonical brand assets from a URL.

## Primary Docs

- Homepage: https://brandkit.danielhowells.com/
- API docs: https://brandkit.danielhowells.com/docs
- Full agent docs: https://brandkit.danielhowells.com/llms-full.txt
- OpenAPI: https://brandkit.danielhowells.com/openapi.json
- MCP: run `pnpm --filter api mcp` from the repository root, or use the checked-in `.mcp.json`.

## Core Endpoints

- `GET /v1/logo?url=example.com` returns a JSON logo response.
- `POST /v1/logo` accepts `{ "url": "example.com", "refresh": false }`.
- `GET /v1/logo.svg?url=example.com` returns an inline SVG logo when available.
- `GET /v1/logo.raster?url=example.com` returns a trimmed raster logo when SVG is unavailable.
- `GET /v1/favicon?url=example.com` returns or redirects to the best favicon.
- `GET /v1/brand?url=example.com` returns a richer brand book.
- `POST /v1/brand` accepts `{ "url": "example.com", "refresh": false }`.
- `GET /v1/logo/recent?limit=24` returns recently fetched logos.

## MCP Tools

- `get_logo`: extracts the best logo and favicon assets for a URL using the framework-free extractor.
- `get_brand_book`: calls the hosted API for the richer brand-book workflow.
- `list_recent_logos`: returns recently cached logo responses from the hosted API.

## Access Policy

- No authentication required.
- Default quota: 60 requests/hour/IP when rate limiting is enabled.
- Add `refresh=1` only when you need a fresh extraction; it bypasses cache and costs more work.
- Results are cached for 30 days.

## Output Notes

- SVGs are sanitised server-side before storage or serving.
- Raster logos are trimmed to remove transparent artboard padding.
- Logo assets include `confidence`, `source`, dimensions, format, and tone metadata when known.
- Missing assets are returned as `null` rather than guessed.