LLM Stack APIs (0.0.1)

Download OpenAPI specification:Download

E-mail: tba@tba.com License: Apache 2.0 Terms of Service

This is a API specs for LLM Stack. You can find out more about LLM Stack at Documentation Page.

Some useful links:

generic

Generic APIs

Executing a flow

Authorizations:

bearerAuth

Request Body schema: application/json

flow_id	string Flow ID
inputs	Array of objects Inputs to the flow, each item is processed independently and results are returned in the same order

Responses

Request samples

Payload

Content type

application/json

{"flow_id": "SGTS-QnA-Bot-1575B4B9",
"inputs": [{"question": "Explain quantum computing in simple terms"
}
]
}

Response samples

200
400

Content type

application/json

{"status": "success",
"data": [{"answer": "Quantum computing is a type of computing that uses the principles of quantum mechanics to process and store information. Unlike classical computers, which use bits to represent information as 0s and 1s, quantum computers use quantum bits, or qubits, which can exist in multiple states at the same time."
}
]
}

Publishing a flow

Publish flow using yaml file.

If flow_id is not defined in the yaml, flow id will be returned to be used in /flows/execute or app.
If flow_id is defined in the yaml, it will update an existing published flow.

Authorizations:

bearerAuth

Request Body schema: multipart/form-data

creator_id	string Creator id to associate with (only required for systems)
flow_yaml_file required	string <binary> Flow config yaml file meeting schema defined at TODO
	object Basic information about the app. If provided, will check if it's an allowed app using yaml's `flow_type`. Required if `flow_type` supports app.

Responses

Response samples

200
400

Content type

application/json

{"status": "string",
"message": "string",
"data": {"api_endpoint": "string",
"flow_id": "string",
"app_url": "string",
"app_settings": {"url_path": "string",
"required_inputs": ["string"
],
"required_outputs": ["string"
]
}
}
}

Get an S3 presigned url to upload file into S3 for flow

Get a presigned URL to upload file into S3 for use in flow execution

query Parameters

flow_id required	string Flow id to associate with file to be stored
filename required	string Filename of file to be stored

Responses

Response samples

200
400

Content type

application/json

{"status": "string",
"data": {"url": "https://s3.ap-southeast-1.amazonaws.com/path",
"s3_object_url": "s3://bucket/path",
"fields": {"key": "string",
"AWSAccessKeyId": "string",
"policy": "string",
"signature": "string"
}
}
}

vsm

Vector Store Management APIs using Execute Flow API

Executing a flow for VSM management - remove `/#vsm` in url

Authorizations:

bearerAuth

Request Body schema: application/json

flow_id	string Flow ID
user_id	string User ID
	Array of objects Inputs to the flow, each item is processed independently and results are returned in the same order

Responses

Request samples

Payload

Content type

application/json

{"flow_id": "Flow_LLMS_Vector_Store_Upsert_Documents_83d6fc4e",
"user_id": "a_b@c.com",
"inputs": [{"vector_store_type": "qdrant",
"collection_name": "my_collection",
"urls": ["https://www.w3.org/WAI/ER/tests/xhtml/testfiles/resources/pdf/dummy.pdf"
],
"file_urls": ["s3://bucket/key"
],
"separators": ["\n",
"\r\n"
],
"chunk_size": 1000,
"chunk_overlap": 20,
"number_of_returned_documents": 9999
}
]
}

Response samples

200
400

Content type

application/json

{"status": "success",
"data": [{"collection_name": "my_collection",
"documents": [{"id": "doc.pdf-1",
"content": "Part 1",
"metadata": {"source": "doc.pdf"
}
},
{"id": "doc.pdf-2",
"content": "Part 2",
"metadata": {"source": "doc.pdf"
}
}
]
}
]
}

Executing a Delete flow for VSM management - remove `/#vsmd` in url

Authorizations:

bearerAuth

Request Body schema: application/json

flow_id	string Flow ID
user_id	string User ID
	Array of objects Inputs to the flow, each item is processed independently and results are returned in the same order

Responses

Request samples

Payload

Content type

application/json

{"flow_id": "Flow_LLMS_Vector_Store_Delete_Documents_ad34ccbb",
"user_id": "a_b@c.com",
"inputs": [{"collection_name": "my_collection",
"document_names": ["test-file.txt"
]
}
]
}

Response samples

200
400

Content type

application/json

{"status": "success",
"data": [{"response": "test-file.txt has been successfully deleted."
}
]
}