Research agents extracting readable text from a known public page
Web intelligence
Structured website intelligence for AI agents.
A website extraction API for AI agents: turn a caller-supplied public URL into clean page text or a structured site report with declared scope and price.
Delx provides governed website extraction for agents that already know which public URL they need to inspect. The current flagship routes return clean page text or a one-call report covering metadata, links, forms, contacts and feed signals. They are extraction tools, not a general web-search engine.
Outcome
A precise job, not a generic promise.
Let an agent inspect a known website without maintaining a scraper, browser pool or separate billing relationship for an occasional task.
Sales and diligence workflows mapping a company's public web surface
Agents checking documentation, contact, form and feed signals in one call
Live capability map
Start from a governed contract.
Each capability links to the property that owns its current terms. Inspect that source immediately before execution.
Page extract
Returns clean title and readable text from one caller-supplied public URL.
Inspect the contract →Website intelligence report
Combines metadata, relevant links, forms, contacts and feed signals for one public site.
Inspect the contract →Domain trust report
A separate governed route for agents evaluating public domain and web-surface signals.
Inspect the contract →Choose the right depth
Page extract vs website intelligence report
Choose page extract when the agent needs clean page text from one known URL. Choose the website intelligence report when it needs a broader site-level view of metadata, links, forms, contacts and feed signals. Both routes require a caller-supplied public URL and currently cost 0.01 USDC per call.
| Criterion | Page extract | Website intelligence report |
|---|---|---|
| Best for | Reading or summarizing one known public page | Mapping the useful public signals around one website |
| Returns | Clean title and readable page text | Metadata, links, forms, contacts and feed signals |
| Route | /api/v1/x402/page-extract | /api/v1/x402/website-intelligence-report |
| Current price | 0.01 USDC per call | 0.01 USDC per call |
Execution
Readable by a person. Runnable by an agent.
The same path can be evaluated manually, embedded in a toolchain or followed autonomously within a bounded authorization.
- Provide a URLSend the exact public page or site the agent already intends to inspect.
- Select depthChoose readable page extraction or a broader one-call website report.
- AuthorizeValidate the 0.01 USDC route price and bounded request before payment.
- Use the resultFeed the structured response into research, diligence or downstream reasoning.
AgentCash · one-call website report
npx agentcash@latest fetch \
https://api.delx.ai/api/v1/x402/website-intelligence-report \
-m POST \
-H 'content-type: application/json' \
-H 'x-delx-source: delx-ai-web-intelligence-page' \
--payment-network base --max-amount 0.01 \
-b '{"url":"https://example.com","timeout":8}'Evidence
What can be verified.
- The extraction pack publishes current routes, prices, example payloads and attribution headers.
- Both flagship operations are independently addressable rather than hidden behind a subscription tier.
- The public discovery copy states the known limitation: general web search remains unavailable.
Boundaries
What is not being claimed.
- The caller must provide a public URL; these routes do not discover arbitrary pages through general search.
- Robots, access controls, timeouts and source-site behavior can limit what a public page returns.
- Extracted content remains subject to the source publisher's rights and the caller's permitted use.
Direct answers
Questions agents and evaluators ask.
Visible answers mirror the structured data on this page.
How can an AI agent extract clean text from a website URL?
Send the known public URL to the page-extract route. It returns the page title and readable text for downstream summarization or analysis; inspect the linked machine contract for the current input and response schema.
Does Delx provide general web search for agents?
No. The current governed offer starts from a public URL supplied by the caller. Delx does not represent extraction as a substitute for a licensed general web-search engine.
What is the difference between page extract and website intelligence?
Page extract focuses on readable content from one page. Website intelligence assembles a broader report of public metadata, links, forms, contacts and feed signals for a site.
How much does website extraction cost?
The two current flagship extraction routes are published at 0.01 USDC per call. The linked machine pack is canonical and should be checked before authorization.