Metadata-Version: 2.4
Name: seekrai
Version: 0.34.4
Summary: Python client for SeekrAI
Author-email: SeekrFlow <support@seekr.com>
License-Expression: Apache-2.0
Project-URL: Homepage, https://www.seekr.com
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.9
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Operating System :: POSIX :: Linux
Requires-Python: <4,>=3.10
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: typer<0.13,>=0.9
Requires-Dist: requests<3.0.0,>=2.31.0
Requires-Dist: tqdm<5.0.0,>=4.66.2
Requires-Dist: tabulate<0.10.0,>=0.9.0
Requires-Dist: pydantic<3.0,>=2.7
Requires-Dist: filelock<4.0.0,>=3.13.1
Requires-Dist: eval-type-backport<0.3.0,>=0.1.3
Requires-Dist: click<9.0.0,>=8.1.7
Requires-Dist: pillow<12.4.0,>=12.3.0
Requires-Dist: pyarrow>=25.0.1
Requires-Dist: numpy>=1.23.5; python_version < "3.12"
Requires-Dist: numpy>=1.26.0; python_version >= "3.12"
Requires-Dist: httpx[http2]<0.28.0,>=0.27.0
Dynamic: license-file

The Seekr Python Library is the official Python client for SeekrFlow's API platform, providing a convenient way for interacting with the REST APIs and enables easy integrations with Python 3.9+ applications with easy to use synchronous and asynchronous clients.

# Installation

To install Seekr Python Library from PyPi, simply run:

```shell Shell
pip install --upgrade seekrai
```

## Setting up API Key

> 🚧 You will need to create an account with [Seekr.com](https://seekr.com/) to obtain a SeekrFlow API Key.

### Setting environment variable

```shell
export SEEKR_API_KEY=xxxxx
```

### Using the client

```python
from seekrai import SeekrFlow

with SeekrFlow(api_key="xxxxx") as client:
    response = client.models.list()
    print(response.data)
```

### Client lifecycle and connection reuse

`SeekrFlow` and `AsyncSeekrFlow` reuse an internal HTTP client per SDK client instance.

Recommended usage:

1. Create one SDK client per app scope or per cached user scope (not per request).
2. Reuse that SDK client for all calls in that scope.
3. Close the SDK client on shutdown/eviction (`close()` or `aclose()`), or use context managers (`with` / `async with`).

If you already manage your own `httpx.Client` / `httpx.AsyncClient` or need custom pool settings, etc, inject it via
`http_client=` or `async_http_client=` and keep lifecycle ownership in your app.

```python
import httpx
from seekrai import SeekrFlow

shared_http_client = httpx.Client(
    http2=True,
    limits=httpx.Limits(max_connections=200, max_keepalive_connections=100),
)
try:
    client = SeekrFlow(api_key="xxxxx", http_client=shared_http_client)
    # ... reuse client across calls
finally:
    shared_http_client.close()
```

#### Per-user SDK cache (recommended pattern)

If your app caches one SDK client per user, use a shared HTTP transport for all
cached SDK instances. This avoids one connection pool per user and keeps
socket/FD usage bounded by one global pool.

```python
import httpx
from cachetools import TTLCache
from seekrai import SeekrFlow

shared_http_client = httpx.Client(
    http2=True,
    limits=httpx.Limits(max_connections=300, max_keepalive_connections=100),
)
sdk_cache: TTLCache[str, SeekrFlow] = TTLCache(maxsize=5000, ttl=900)


def get_user_client(user_id: str, api_key: str) -> SeekrFlow:
    client = sdk_cache.get(user_id)
    if client is None:
        client = SeekrFlow(api_key=api_key, http_client=shared_http_client)
        sdk_cache[user_id] = client
    return client


def shutdown() -> None:
    # SDK clients do not own injected transport.
    shared_http_client.close()
```

Apply the same pattern for async apps with one shared `httpx.AsyncClient` injected
into cached `AsyncSeekrFlow` instances, and close it once during app shutdown.

### RBAC Team Routing

Every API key currently resolves to a personal team. We do not currently have
application-level or shared team-level API keys.

The SDK can send the RBAC context header `x-team-id` when provided. This header
does not grant access by itself.

Authorization is enforced server-side using the authenticated identity
(API key/JWT). A request is only allowed if that identity has access to the
requested team. If no team context is provided, the backend defaults to the
personal team resolved from authentication.

You can set team context in either of these ways:

1. Set `SEEKR_TEAM_ID` and let the SDK populate `x-team-id` automatically.
2. Pass `supplied_headers={"x-team-id": "..."}` explicitly.

If both are provided, `supplied_headers["x-team-id"]` takes precedence.

```python
import os
from seekrai import SeekrFlow

with SeekrFlow(api_key=os.environ.get("SEEKR_API_KEY")) as client:
    response = client.models.list()
    print(response.data)
```

```python
import os
import asyncio
from seekrai import AsyncSeekrFlow

async def run():
    async with AsyncSeekrFlow(
        api_key=os.environ.get("SEEKR_API_KEY"),
        supplied_headers={"x-team-id": os.environ.get("SEEKR_TEAM_ID")},
    ) as async_client:
        response = await async_client.models.list()
        print(response.data)

asyncio.run(run())
```

# Usage – Python Client

## Chat Completions

```python
import os
from seekrai import SeekrFlow

with SeekrFlow(api_key=os.environ.get("SEEKR_API_KEY")) as client:
    response = client.chat.completions.create(
        model="meta-llama/Llama-3.1-8B-Instruct",
        messages=[{"role": "user", "content": "tell me about new york"}],
    )

print(response.choices[0].message.content)
```

### Streaming

```python
import os
from seekrai import SeekrFlow

with SeekrFlow(api_key=os.environ.get("SEEKR_API_KEY")) as client:
    stream = client.chat.completions.create(
        model="meta-llama/Llama-3.1-8B-Instruct",
        messages=[{"role": "user", "content": "tell me about new york"}],
        stream=True,
    )

    for chunk in stream:
        print(chunk.choices[0].delta.content or "", end="", flush=True)
```

### Async usage

```python
import os, asyncio
from seekrai import AsyncSeekrFlow

messages = [
    "What are the top things to do in San Francisco?",
    "What country is Paris in?",
]


async def async_chat_completion(messages):
    async with AsyncSeekrFlow(api_key=os.environ.get("SEEKR_API_KEY")) as async_client:
        tasks = [
            async_client.chat.completions.create(
                model="meta-llama/Llama-3.1-8B-Instruct",
                messages=[{"role": "user", "content": message}],
            )
            for message in messages
        ]
        responses = await asyncio.gather(*tasks)

        for response in responses:
            print(response.choices[0].message.content)


asyncio.run(async_chat_completion(messages))
```

## Files

The files API is used for fine-tuning and allows developers to upload data to fine-tune on. It also has several methods to list all files, retrieve files, and delete files

```python
import os
from seekrai import SeekrFlow

with SeekrFlow(api_key=os.environ.get("SEEKR_API_KEY")) as client:
    client.files.upload(file="somedata.parquet")  # uploads a file
    client.files.list()  # lists all uploaded files
    client.files.delete(id="file-d0d318cb-b7d9-493a-bd70-1cfe089d3815")  # deletes a file
```

## Fine-tunes

The finetune API is used for fine-tuning and allows developers to create finetuning jobs. It also has several methods to list all jobs, retrieve statuses and get checkpoints.

```python
import os
from seekrai import SeekrFlow

with SeekrFlow(api_key=os.environ.get("SEEKR_API_KEY")) as client:
    client.fine_tuning.create(
        training_file='file-d0d318cb-b7d9-493a-bd70-1cfe089d3815',
        model='meta-llama/Llama-3.1-8B-Instruct',
        n_epochs=3,
        n_checkpoints=1,
        batch_size=4,
        learning_rate=1e-5,
        suffix='my-demo-finetune',
    )
    client.fine_tuning.list()  # lists all fine-tuned jobs
    client.fine_tuning.retrieve(id="ft-c66a5c18-1d6d-43c9-94bd-32d756425b4b")  # retrieves information on finetune event
```
