# 2025.34 Public API / SDK Changes

## Public API

### **New Endpoints**

tech`DELETE` **Delete a Chat** - API to delete a chat.

**Example**

```
curl --location --request DELETE 'https://gateway.oleole.unique.app/public/chat/space/chat/chat_se62mrxrt4k0rksimr1ojzmv' 
--header 'x-api-version: 2023-12-06' 
--header 'x-user-id: <your-user-id>' 
--header 'x-company-id: <your-company-id>' 
--header 'Authorization: Bearer <your-api-key>' 
--header 'x-app-id: <your-app-id>' 
```

tech`POST` **Stream to the chat using the Responses API the Responses API** - Streams the answer to the chat frontend using the Responses API.

**Example**

```
curl --location 'https://gateway.oleole.unique.app/public/chat/integrated/chat/stream-responses' 
--header 'x-api-version: 2023-12-06' 
--header 'x-user-id: <your-user-id>' 
--header 'x-company-id: <your-company-id>' 
--header 'Authorization: Bearer <your-api-key>' 
--header 'x-app-id: <your-app-id>' 
--header 'Content-Type: application/json' 
--data '{
    "chatId": "chat_vcrctshcn0wj5qxfqafwr83h",
    "model": "AZURE_o3_2025_0416",
    "assistantMessageId": "msg_k27vdzd6dpc3ys846za3s2i9",
    "userMessageId": "msg_g0yub3u2g7ab0azpcukmeg4x",
    "input": "Tell me about the curious case of neural text degeneration",
    "reasoning": {
        "effort": "medium",
        "summary": "detailed"
    }
}'
```

## SDK

tech **Chat Against a File** - The script creates a chat and uploads the file to it. It then keeps polling the `ingestionState` field of the message, waiting for it to reach `FINISHED`, signaling the upload is complete. Once the file uploads successfully, the script sends the text, continues polling for completion, and finally retrieves the response message.

You must provide the following parameters:

- `assistantId`: The assistant to be used for the chat.
- `path_to_file`: The local path of the file to be uploaded.
- `displayed_filename`: The name of the file to be displayed.
- `mime_type`: The mime type of the file to be uploaded.
- `text`: The text to be sent to the chat for chatting against the file.

**Optional parameters:**

- `poll_interval`: The number of seconds to wait between polling attempts (default: `1` second).
- `max_wait`: The maximum number of seconds to wait for the message to complete (default: `60` seconds).

Example of chatting against a PDF (the usage can be extended to any supported file type)

py

```
latest_message = await unique_sdk.utils.chat_in_space.chat_against_file(
    user_id=user_id,
    company_id=company_id,
    assistant_id="assistant_hjcdga64bkcjnhu4",
    path_to_file="/files/hello.pdf",
    displayed_filename="hello.pdf",
    mime_type="application/pdf",
    text="Give me a bullet point summary of the file.",
)
```

tech **Delete Chat** - Delete a space chat by id. If the chat does not exist, the function will return an error.

**Example**

py

```
unique_sdk.Space.delete_chat(
    user_id=user_id,
    company_id=company_id,
    chat_id="chat_dejfhe729br398",
)
```

tech **Stream Using Responses API** - Streams the answer to the chat frontend using the Responses API.

**Example**

py

```
unique_sdk.Integrated.responses_stream(
    user_id=userId,
    company_id=companyId,
    model="AZURE_o3_2025_0416",
    assistantMessageId=assistantMessageId,
    userMessageId=userMessageId,
    input="Tell me about the curious case of neural text degeneration",
    chatId=chatId,
)
```

tech **Get Paginated Folders** - Get paginated folders info based on parentId. If the parentId is not defined, the root folders will be returned.

py

```
unique_sdk.Folder.get_infos(
    user_id=user_id,
    company_id=company_id,
    take=10,                                    #optional
    skip=5,                                     #optional
    parentId="scope_s18seqpnltf35niydg77xgyp"   #optional
)
```

tech **Get Paginated Contents** - Get contents by using either a parent id or a metadata filter in UniqueQL language. If both are defined, the function will throw an error.

**Example of metadata filter defined in UniqueQL:**

py

```
metadataFilter: {
    "or": [\
        {\
            "and": [\
                {\
                    "operator": "contains",\
                    "path": [\
                        "folderIdPath"\
                    ],\
                    "value": "uniquepathid://test_id"\
                },\
                {\
                    "operator": "contains",\
                    "path": [\
                        "title"\
                    ],\
                    "value": "ai"\
                }\
            ]\
        }\
    ]
},
```

Pagination is also enabled for this functionality, and the default number of returned results is 50 with no entries skipped. Use the following parameters to get the desired page:

- `skip`
- `take`

Here is an example of retrieving the first 3 content infos that are in the `scope_abcdibgznc4bkdcx120zm5d` scope and have the value `ai` in the `title`.

py

```
content_info_result = unique_sdk.Content.get_infos(
    user_id=user_id,
    company_id=company_id,
    metadataFilter={
        "or": [\
            {\
                "and": [\
                    {\
                        "operator": "contains",\
                        "path": [\
                            "folderIdPath"\
                        ],\
                        "value": "scope_abcdibgznc4bkdcx120zm5d"\
                    },\
                    {\
                        "operator": "contains",\
                        "path": [\
                            "title"\
                        ],\
                        "value": "ai"\
                    }\
                ]\
            }\
        ]
    },
    skip=0,
    take=3,
)
```

Here is an example of retrieving the contents based on a parentId.

py

```
content_info_result = unique_sdk.Content.get_infos(
    user_id=user_id,
    company_id=company_id,
    parentId="scope_ahefgj389srjbfejkkk98u"
)
```

| **Author** |  |