Search for agents.
Structured results. No ads. No SEO spam. Just facts, snippets, and citations. x402 auth.
# Search the web $ curl -G https://seek.prim.sh/v1/search \\ -H "X-402-Payment: $TOKEN" \\ --data-urlencode "q=best GPU for LLM inference 2026" { "results": [ { "title": "H100 vs RTX 5090 Benchmarks", "url": "https://benchmarks.ai/...", "snippet": "The RTX 5090 offers 70% of H100 performance...", "published": "2026-01-15" } ], "answer": "For inference, RTX 5090 (32GB) is the best value..." } # Crawl a page (markdown conversion) $ curl -G https://seek.prim.sh/v1/extract \\ -H "X-402-Payment: $TOKEN" \\ --data-urlencode "url=https://benchmarks.ai/gpu-review"
x402 auth
No ads
Markdown extraction
Fact checking
Part of agentstack

What agents use it for

Research & fact-checking

Agent needs to verify a claim or find technical specs. seek.sh provides authoritative sources without the SEO fluff.

Content extraction

Turn any URL into clean Markdown. Strips ads, navbars, popups. Just the content the agent needs to read.

News monitoring

"Find latest CVEs for Nginx." Agent runs a daily search, diffs results, and alerts on new vulnerabilities.

Competitor analysis

Agent scans pricing pages of 5 competitors. extract endpoint turns HTML tables into JSON data.

API reference

GET    /v1/search               # Web search
GET    /v1/search/news          # News specific search
GET    /v1/extract              # URL → Markdown
GET    /v1/screenshot           # URL → Image (for vision models)

Pricing

ActionCostNotes
Search query$0.002Per request
Extraction$0.001Per URL
Screenshot$0.005Full page render

The web is an API call.

Agents don't browse. They index.

Read the docs →