Agent Seek docs

Agent Seek is agent search: cheap You.com recall + TypeSafe Jev taste → ranked web results. Google-simple for people. Judgment-native for agents.

What Agent Seek is

Agent Seek is agent search for humans and coding agents: You.com discovers web candidates, TypeSafe Jev cascade-ranks them, and you get ranked web results — title, URL, snippet, relevance score, flags, and signals — instead of raw SERP mush.

Who it is for

People use the Google-simple box on the search UI (type a query, press Enter). Agents call the Agent Seek API (POST /v1/search) with OAuth 2.0 (search:read) or Authorization: Bearer $AGENT_SEEK_API_KEY. Use Agent Seek when you need a small set of high-relevance sources for research or RAG and you will write the answer yourself. Do not use it as a chat model, an unbounded crawler, or a generative engine. Hard cap: 100 discover candidates per query (default 50 in, top 10 out).

Auth is OAuth 2.0 plus a shared API-key alternate. The prototype is free. See also authentication, MCP Streamable HTTP, llms.txt, about, and the Agent Seek skill. OpenAPI: /openapi.json (Swagger /api/docs).

Install the skill

Efficient path: skill → /llms.txt → /auth.md → MCP or REST.

• Stable skill: /.well-known/agent-skills/agent-seek/SKILL.md (also /skills/agent-seek/SKILL.md)
• Index: /.well-known/agent-skills/index.json
• Cursor: save SKILL.md as .cursor/skills/agent-seek/SKILL.md (project) or ~/.cursor/skills/agent-seek/SKILL.md (user).
• Other coding agents: download SKILL.md from the well-known URL, or link it from the index.

Then read /llms.txt, authenticate via /auth.md, and call POST /v1/search or MCP (/mcp, server card).

Humans

• Open the search UI — type a query and press Enter.
• Each result shows a relevance meter; hover for a signal breakdown (UI label for prompt_injection is Injection risk).
• UI and API both default to deep mode (k=10). No Mode/Order controls. Pass mode=snip on the API for the cheaper path.

Agent Seek API

The crawlable developer entry is /developers (title: Agent Seek API). Contract: POST /v1/search. Schema: /openapi.json. Swagger: /api/docs. Auth: /auth.md.

Agents (primary)

Live prototype: https://agentseek.dev (alternate: https://gitmaxd-agent-seek.exe.xyz)

Public HTTP API. Agents use OAuth 2.0 (search:read) or AGENT_SEEK_API_KEY (see auth.md). MCP: /mcp.

Auth header:

Authorization: Bearer $AGENT_SEEK_API_KEY

Contract: POST /v1/search

{
  "q": "your query",
  "k": 10,
  "max_candidates": 50,
  "mode": "deep",
  "nocache": false
}

• mode: deep (default — fetch survivor pages, then rank) or snip (title/URL/snippet, cheaper).
• k 1–25 (default 10) · max_candidates 1–100 (default 50).
• Results include score, flags, and optional signals: answerability, authority, on_topic, states_sought_fact, subject_match, spam, prompt_injection.
• Hard gates (order, fail-open if a parse is missing): subject_match < 0.55, then is_republisher ≥ 0.55, then prompt_injection ≥ 0.55 exclude from results (still in raw_results). UI label: Injection risk.

When to use: need a small set of high-relevance URLs/snippets for research or RAG.

When not: generative answers, unbounded crawl, or >100 candidates.

Curl:

curl -sS -X POST "https://agentseek.dev/v1/search" \
  -H "Authorization: Bearer $AGENT_SEEK_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"q":"Introducing System One Models Jev","k":10,"mode":"deep"}'

Also

• GET /health — liveness + version
• Agent Seek sandbox — zero-auth canned SearchResponse
• Agent Seek OpenAPI · Swagger · ReDoc
• Agent Seek MCP · server card