📙Reading & writing data via the API

Glif API is in beta and subject to change.

For fetching info about a specific workflow, user, or a specific glifRun.

Paginate using ?page= when applicable. You can also increase the # of results per page using e.g. ?limit=50, currently defaulting to 20 and capped at 100

Individual workflow

Fetch a specific workflow: GET https://glif.app/api/glifs?id=clgh1vxtu0011mo081dplq3xs

Includes full glif-graph JSON in the data field

Lists of workflows

Fetch all public workflows: GET https://glif.app/api/glifs

Fetch featured workflows on homepage: GET https://glif.app/api/glifs?featured=1

Fetch user's workflows by username: GET https://glif.app/api/glifs?username=jamiedubs

Fetch user's workflows by user ID: GET https://glif.app/api/glifs?userId=clcu26uxx0000lc08mqnjf5mq

To get the full glif-graph JSON in the data field, append ?includes=spells.data

Runs for a given workflow

Fetch a specific run: GET https://glif.app/api/runs?id=i3kqrilvp6rmsszwnw44ycyl

Fetch all public runs for a specific workflow: GET https://glif.app/api/runs?glifId=clgh1vxtu0011mo081dplq3xs (?spellId= also works)

Fetch all public runs by a specific user:

Create a workflow

Currently in private testing; contact us if you're messing with this!

Here's a curl example which creates an empty, unpublished (draft) workflow:

curl -X POST "https://glif.app/api/glifs" \
     -H "Authorization: Bearer $GLIF_API_TOKEN" \
     -H "Content-Type: application/json" \
     -d '{
       "name": "API test glif",
       "description": "just testing :)",
       "graph": {"nodes": []}
     }'

Here's a curl example that includes one text input and an LLM call (GPTBlock) copied from this Perplexity search workflow.

Please note that formatting complex JSON inside a shell command like this is a huge pain in the butt and is not recommended for serious development :)

curl -X POST "https://glif.app/api/glifs" \
  -H "Authorization: Bearer $GLIF_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "API test glif 2",
    "description": "just testing again, I love APIs",
    "graph": {
      "nodes": [
        {
          "name": "query",
          "type": "TextInputBlock",
          "params": {
            "label": "query",
            "value": "what is trending in pop culture this week?"
          }
        },
        {
          "name": "result",
          "type": "GPTBlock",
          "params": {
            "model": "Perplexity: Llama3 Sonar 8B Online",
            "prompt": "{query}",
            "jsonMode": false,
            "maxTokens": 1000,
            "temperature": 0.2,
            "systemPrompt": "You are a helpful assistant."
          }
        }
      ]
    }
  }'

Here's an example response on success for that call:

{"id":"clzs7eoan0003n1xgod8grhaj","name":"API test glif 2","imageUrl":null,"description":"just testing again, I love APIs","createdAt":"2024-08-13T09:12:02.688Z","updatedAt":"2024-08-13T09:12:02.688Z","publishedAt":null,"output":null,"outputType":null,"forkedFromId":null,"featuredAt":null,"userId":"clcu26uxx0000lc08mqnjf5mq","completedSpellRunCount":0,"averageDuration":null,"user":{"id":"clcu26uxx0000lc08mqnjf5mq","name":"jamiedubs","image":"https://res.cloudinary.com/dzkwltgyd/image/upload/v1709593997/user-image-production/ckkbd7jmxly9ms0csaxv.jpg","username":"jamiedubs"},"_count":{"likes":0,"comments":0},"spheres":[],"data":{"nodes":[{"name":"query","type":"TextInputBlock","params":{"label":"query","value":"what is trending in pop culture this week?"}},{"name":"result","type":"GPTBlock","params":{"model":"Perplexity: Llama3 Sonar 8B Online","prompt":"{query}","jsonMode":false,"maxTokens":1000,"temperature":0.2,"systemPrompt":"You are a helpful assistant."}}]}}

Note the glif id, clzs7eoan0003n1xgod8grhaj. Also note that this workflow is also unpublished and so it is not accessible to any user except you! You can publish it via the website, or by specifying a "publishedAt" date in your POST request.

You can then run it this workflow:

curl -X POST "https://simple-api.glif.app" \
  -H "Authorization: Bearer $GLIF_API_TOKEN" \
  -d '{"id": "clzs7eoan0003n1xgod8grhaj", "query": "who is the coolest software developer of all time?"'}

To view raw workflow data, you can fork or build something on glif.app and use the "Debug" button bottom-left.

Alternately, it's present in all published workflow via the API, e.g., in order of increasing complexity:

User info

Fetch info about the logged-in user (and thus test that your API token is working and you are logged-in): GET https://glif.app/api/me

Fetch info about a user, via either ID or username:

Spheres (beta)

Spheres are basically our staff-curated collections or categories.

To list all public spheres: GET https://glif.app/api/spheres

To get workflows from a specific sphere (AKA folder or collection): GET https://glif.app/api/spheres?slug=memes

You can also use ?id= param

Like the 'list of workflows' above, you can fetch the full glif-graph JSON in these list queries by appending ?includes=spells.data

Public spheres are listed on https://glif.app/spheres

Help/Support/Questions

Please join our Discord and post in the #get-help channel if you have any questions or need help with the API. We'd love to hear from you!

Alternately you can contact us via email, but those will get slower replies.

PreviousUsing AI Assistants to build with the Glif APINextGlif Graph JSON Schema

Last updated