Tasklin

Tasklin is a Python CLI for integrating multiple AI providers into scripts, pipelines, or automation workflows. It’s built for developers who want structured AI outputs for tasks, processing, or data generation, without being limited to interactive chat.

Website: https://tasklin.dev

Features

• Single CLI for multiple AI providers
• Supports OpenAI, Ollama, Anthropic (Claude), DeepSeek, and more
• Returns structured responses including tokens used and execution time
• Supports sync and async execution
• Clean error handling for missing models, invalid API keys, etc.
• Cross-platform
• Installable from PyPI as tasklin

Installation

Clone the repo and install dependencies:

git clone https://github.com/jetroni/tasklin.git
cd tasklin
python -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install -r requirements.txt

Or install directly from PyPI:

pip install tasklin

(Optional) Make the CLI globally executable:

pip install --editable .

Usage

Basic example:

tasklin "Generate a report from this data" --type openai --key MY_KEY

Ollama example (self-hosted):

tasklin "Run analysis" --type ollama --base-url http://localhost:11434 --model codellama

⚠️ Async mode is planned but not implemented yet.

CLI Options

Response Structure

All responses are returned as a structured JSON object:

{
  "provider": "openai",
  "model": "gpt-4o-mini",
  "output": "Task output here",
  "raw": {...},
  "tokens_used": 17,
  "duration_ms": 543
}

Contributing

Contributions are welcome:

• Add new AI providers
• Improve error handling
• Enhance async execution and CLI experience

Fork the repo and submit a pull request.

PyPI

Install via:

pip install tasklin

Latest releases: https://pypi.org/project/tasklin/