Skip to main content
Search backend in an AgentField example AgentField’s deep research agent breaks a research question into subtasks, searches the web for each leaf task and writes a report. Set SEARCH_PROVIDER=keenable and its searches go to Keenable. Tavily stays the default. Keenable needs no account or API key. → Deep research agent README

Install

The example lives in the AgentField repository. From a clone, install its requirements:

Run the agent on Keenable

Select Keenable and start the agent:
In a second terminal, send a research question:
The agent plans its subtasks, runs one Keenable search per leaf task and returns the synthesized report.

Settings

Without a key, searches run on the shared public tier at lower rate limits. A deep research run makes many searches, so export KEENABLE_API_KEY for larger questions. See API keys.

npcpy

Search the web from npcpy with the keenable provider for search_web, or call search_keenable directly. No API key needed, in npcpy 2.1.28 and later.

RL and eval environments

Use Keenable as the search backend in SkillOpt OfficeQA rollouts and Halo GRPO search environments: one config value, opt-in, never a default.

Vercel AI SDK

Keyless web search and page fetch for the Vercel AI SDK (@keenable/ai-sdk), plus Vercel Connect to keep the API key out of environment variables.

Tool calling

Give an LLM web search and page fetch as tools. Pass the SDK’s TOOLS to any OpenAI-compatible chat completions client and run each call with run_tool_call.