AICP-EAT

Eat Every Python Library → HTTP API.

pip install aicp-eat

Eat whatever you want

pip install pandas
python eater/eat.py pandas            # Standard library — feed it to eat.py

pip install numpy
python eater/eat.py numpy             # Standard library — feed it to eat.py

pip install pillow
python eater/eat_pil.py               # Nested modules — throw it to AI, let it generate a custom script

pip install langchain-community
python eater/eat_langchain.py         # Nested modules — throw it to AI, let it generate a custom script

plugins/
├── pandas.py              → 102 functions
├── numpy.py               → 460 functions
├── PIL.py                 → 664 functions
└── langchain_community.py → 7512 functions
────────────────────────────────────────
Total: 8738 functions. 4 files.

and more. Come eat.

Run

# Set your API key in aicp.yaml, or:
export AGGREGATOR_API_KEY="sk-your-key"  # Windows: setx AGGREGATOR_API_KEY "sk-your-key"

aicp

curl -X POST http://localhost:9000/api/pandas \
  -H "Content-Type: application/json" \
  -d '{"function":"array","args":{"data":[1,2,3,4,5]}}'

Return

{
  "result": "IntegerArray\n[1, 2, 3, 4, 5]\nLength: 5, dtype: Int64",
  "source": "pandas"
}

Open http://localhost:9000, browse the API menu at /api/pandas, /api/numpy, /api/PIL, /api/langchain_community. Each one has a help() with all functions listed.

Why eat the entire Python ecosystem? →

Go calls pandas. Rust calls numpy. Frontend calls LangChain. Any language, HTTP everything.

AI reads help(). AI orchestrates. 80 lines engine.AI-Generated Experiments

AI read the AICP protocol and generated complete systems across seven domains. Each from a single human sentence.

→ More about the protocol

Dependencies

• Python >= 3.10
• aiohttp >= 3.9
• openai >= 1.0
• pyyaml >= 6.0

Eat it, Eat it, Eat it, Eat it
No one wants to be defeated
Showin' how funky and strong is your fight
It doesn't matter who's wrong or right
Just eat it, eat it
Just eat it, eat it

See More →

MIT License · Dvwoo