Skip to main content

APIs

Promptly provides a set of APIs that you can use to access the functionality of Promptly. You can use these APIs to run apps, manage your data sources etc.,

Authentication​

All Promptly APIs require authentication. You can authenticate using your API Token. You can find your API Token in your Promptly Settings. You can pass your API Token in the Authorization header with Token prefix.

Streaming Output​

For app runs, if stream parameter is used, output is received in chunks of values for JSON keys. For example, if the output is a JSON object with keys a, b and c, you will receive the output in the following format:

{
  "a": <A_VALUE_CHUNK>,
  "b": <B_VALUE_CHUNK>,
  "c": <C_VALUE_CHUNK>
}

You will need to parse the output in your client code and stitch the chunks together to get the complete output.

API Reference​

Apps​

Run an app.​

POST /api/apps/<app_uuid>/run

or

POST /api/apps/<app_uuid>/run/<session_id>

where session_id is a unique identifier for the session. You can use the same session id to run the app multiple times. If you don't pass a session id, Promptly will generate a random session id for you.

Request body​

You can pass the values for the app's input variables in the request body. You can also pass stream parameter to stream the response.

{
  "input": <KEY_VALUE_JSON>,
  "stream": <BOOLEAN>
}
Response body​
{
  "session": {
    "id": <SESSION_ID>,
  },
  "output": <APP_OUTPUT>
}

Create an App​

You can create an app by sending a POST request to /api/apps with the app definition as YAML in the request body. Make sure to set the Content-Type header to application/yaml.

POST /api/apps

Update an App​

You can update an app by sending a PATCH request to /api/apps/<app_uuid> with the app definition as YAML in the request body. Make sure to set the Content-Type header to application/yaml.

PATCH /api/apps/<app_uuid>

Datasources​

List datasources​

GET /api/datasources
Response body​
[
  {
    "uuid": <DATASOURCE_UUID>,
    "name": <DATASOURCE_NAME>,
    "type": <DATASOURCE_TYPE>,
    "created_at": <DATASOURCE_CREATED_AT>,
    "updated_at": <DATASOURCE_UPDATED_AT>,
  },
  ...
]

Create a datasource​

POST /api/datasources
Request body​
{
  "name": <DATASOURCE_NAME>,
  "type": <DATASOURCE_TYPE>,
}
Response body​
{
  "uuid": <DATASOURCE_UUID>,
  "name": <DATASOURCE_NAME>,
  "type": <DATASOURCE_TYPE>,
  "created_at": <DATASOURCE_CREATED_AT>,
  "updated_at": <DATASOURCE_UPDATED_AT>,
}

Delete a datasource​

DELETE /api/datasources/<datasource_uuid>

Add entry to a datasource​

POST /api/datasources/<datasource_uuid>/entries
Request body​
{
  "entry_data": {
    "name": "<ENTRY_NAME>",
    "content": "<ENTRY_CONTENT>",
    "url": "<ENTRY_URL>",
    "file": "<BASE64_ENCODED_FILE_DATA>"
  },
  "entry_metadata": {
    "key1": "value1",
    "key2": "value2"
  }
}

content, url and file parameters are optional. You can pass any one of them based on the type of datasource. entry_metadata is an optional key-value JSON object that can be used to store additional metadata about the entry. It is used when you want to filter entries based on metadata during vector search.

note

Metadata keys will be available with md_ prefix in the vector store. For example, if you have a metadata key key1, it will be available as md_key1 in the vector store. There is default metadata like source available on all entries and is auto generated based on the entry.

Delete entry from a datasource​

DELETE /api/datasources/<datasource_uuid>/entries/<entry_uuid>

List entries in a datasource​

GET /api/datasources/<datasource_uuid>/entries