> ## Documentation Index
> Fetch the complete documentation index at: https://docs.scitix.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# wiseflow

> Learn how to configure LLM_API_BASE in wiseflow and use ScitiX models for information mining and intelligence aggregation tasks.

[wiseflow](https://github.com/TeamWiseFlow/wiseflow) is an open-source information mining and intelligence aggregation tool that extracts useful information from web pages and other sources based on your focus areas. It calls large language models through the native OpenAI SDK, so you can connect wiseflow to ScitiX Model Inference.

This guide uses `glm-5.2` as an example. For the full model list and pricing, see [Models](https://console.scitix.ai/model-inference/models).

## Configure

Create an API Key on the [API Keys](https://console.scitix.ai/model-inference/api_keys) page. In the project root, edit `.env`:

```bash theme={null}
LLM_API_BASE="https://api.scitix.ai/model-api/v1"
LLM_API_KEY="<YOUR_API_KEY>"
PRIMARY_MODEL="glm-5.2"
SECONDARY_MODEL="glm-5.2"
```

| Field             | Description                                                                                     |
| ----------------- | ----------------------------------------------------------------------------------------------- |
| `LLM_API_BASE`    | ScitiX OpenAI-compatible base URL with `/v1`: `https://api.scitix.ai/model-api/v1`              |
| `LLM_API_KEY`     | Your API Key created on the [API Keys](https://console.scitix.ai/model-inference/api_keys) page |
| `PRIMARY_MODEL`   | Primary model ID, such as `glm-5.2`                                                             |
| `SECONDARY_MODEL` | Secondary model for initial filtering. It can be the same as the primary model.                 |

## Verify

wiseflow reads the configuration above in `core/llms/openai_wrapper.py` and calls the model through the native OpenAI SDK. After you start a task, model requests are sent to the ScitiX endpoint.

You can also run a minimal check:

```python theme={null}
import asyncio
from core.llms.openai_wrapper import openai_llm

print(asyncio.run(openai_llm(
    [{"role": "user", "content": "Reply with exactly: OK"}],
    model="glm-5.2",
)))
```

For normal operation, run `python core/run_task.py`. Extracted information is written to PocketBase by default at `http://127.0.0.1:8090/_/`.

## FAQ

* **Version**:
  Use a 0.3.x release, such as tag `v0.3.8`. The repository's `master` branch has moved to a different product, while the classic Python + PocketBase implementation is available in versioned tags.
* **`LLM_API_BASE`**:
  Use the base URL with `/v1`: `https://api.scitix.ai/model-api/v1`. wiseflow uses the native OpenAI SDK and does not require a dedicated provider.
* **Model IDs**:
  Use model IDs from the [Models](https://console.scitix.ai/model-inference/models) page.