Keenable SDK

Official SDKs for Keenable, a web search API built for AI agents. Search returns ranked pages with their text already extracted, so one call gives a model something to reason over; fetch returns any URL as clean markdown.

Keyless by default. Every call works with no account and no key. An optional KEENABLE_API_KEY only lifts the hourly rate limit.

Language	Package	Source
Python 3.9+	keenable	python/
TypeScript / Node 18+	keenable	typescript/

pip install keenable
npm install keenable

from keenable import Keenable

keenable = Keenable()
results = keenable.search("fastest inference providers 2026")
print(results.to_context())   # numbered, citable block, ready for a prompt

import { Keenable } from "keenable";

const keenable = new Keenable();
const results = await keenable.search("fastest inference providers 2026");
console.log(results.toContext());

Both SDKs expose the same surface: search, fetch, a to_context() / toContext() renderer for prompts, and OpenAI-compatible tool definitions for tool calling. See the per-language READMEs for the full reference.

Examples

• examples/cerebras/ - grounding a Cerebras model in live web results, and letting it call Keenable as a tool. Python and Node.

Documentation for partners

• docs/partners/cerebras/keenable.mdx - integration page written for the Cerebras Inference docs.

License

MIT