Dog API

A free JSON API about dogs: 283 breeds, 9 breed groups, 483 facts and about 2,350 openly licensed breed photos. No API key, no sign-up, CORS open. Running at dogapi.dog since 2016.

Quick start

curl "https://dogapi.dog/api/v2/facts?limit=2"
curl "https://dogapi.dog/api/v2/breeds?page[size]=5"
curl "https://dogapi.dog/api/v2/groups"

Responses follow JSON:API. fetch() works straight from a browser, a CodePen, a notebook or localhost; no proxy needed. The limit is 300 requests per minute per IP.

A random dog in one tag

<img src="https://dogapi.dog/api/v2/breeds/image" alt="A random dog">

The URL redirects to a picture. Add ?size=thumb|medium|large|full, or use /api/v2/breeds/{id}/image for a specific breed. Photos come from Wikimedia Commons and Openverse; each breed's attribution is in the API response.

Use it from an AI assistant (MCP)

A remote MCP server lives at https://dogapi.dog/mcp, no key needed. Tools: random_dog_facts, search_breeds, get_breed, list_groups, get_group, random_dog_image.

claude mcp add --transport http dogapi https://dogapi.dog/mcp

Or in any client that takes JSON config:

{ "mcpServers": { "dogapi": { "type": "http", "url": "https://dogapi.dog/mcp" } } }

Docs

• API v2 reference, with an OpenAPI document for Postman, Insomnia or code generators
• API v1 still answers, so old tutorials keep working

Used in classrooms

A lot of the traffic comes from students: assignments on Canvas, Google Classroom, Moodle and Schoology, Colab notebooks and The Odin Project. If you teach, dogapi.dog/teach has ready-made exercises with starter code in JavaScript, Python and curl. Free for classes, workshops, tutorials and videos, paid or not.

Running it locally

You need Ruby, PostgreSQL and libvips (for image variants).

1. Clone this repository
2. bundle install
3. rails db:setup
4. rails server

Breed images

Breed pictures are imported into Active Storage, resized to WebP, and served with the attribution their licence requires. libvips must be installed locally for the variants.

Five sources are tried in turn, in order of how much human judgement went into what they return: the photo a breed's Wikipedia article leads with, then every other picture that article and its translations use, then the files in its Wikimedia Commons category, then a Commons search, then the openly licensed photographs Openverse indexes.

rails "images:backfill[10]"      # walk every source until each breed has ten
rails "images:import[Akita]"     # one breed, one source
rails images:stats               # coverage so far
rails images:reprocess           # rebuild missing variants

Both the importing and the scoring are mostly spent waiting on somebody else's API, so long runs belong on Sidekiq rather than in a terminal:

rails "images:backfill_async[10]"   # a job per breed, chained through the sources
rails images:rerank_async           # a job per unscored picture

Two workers handle them, because Wikimedia's patience and the model's rate limit are nothing alike — see docker-compose.yml:

bundle exec sidekiq -q images -c 2
bundle exec sidekiq -q reviews -q default -c 8

A breed's first image is the one the API and the site show on their own, so a review pass looks at each picture and scores it as the one photograph a breed page leads with: one adult dog, sharp, filling the frame, nothing else in it. The best scoring picture is moved to the front, and the score and the sentence behind it are kept on the record. A score of 0 means the wrong picture rather than a poor one: no dog in it, or several photographs arranged into a grid. This needs ANTHROPIC_API_KEY.

rails "images:rerank[Akita]"     # score one breed's pictures and reorder them
rails images:rerank_all          # every breed, skipping pictures already scored
rails images:scores              # how the scores came out
rails "images:prune_reviewed[4]" # delete everything that scored below 4/10

Hand picked images can be listed in db/seeds/breed_images.yml and imported with rails "images:import[Akita,manual]".

In production set ACTIVE_STORAGE_SERVICE to amazon or cloudflare and fill in the S3_* values from .env.example; without them files stay on local disk.

Contributing

We welcome contributions to this project! If you have an idea for a new feature or find a bug, please open an issue in this repository.

To contribute code to the project, follow these steps:

1. Fork this repository
2. Create a new branch for your changes
3. Make the necessary changes and commit them to your branch
4. Push your branch to your forked repository
5. Open a pull request from your branch to this repository

We will review your changes and merge them into the project if they are approved.

License

This project is licensed under the MIT License. See the LICENSE file for details.