jsonrecon

JSON Recon Full Schema Extraction converts a public webpage into concise, validated Schema.org JSON-LD built for AI agents, autonomous workflows, search tools, data pipelines, RAG systems, MCP clients, and application developers. Submit a URL and receive a clean machine-readable entity instead of spending model context on navigation, boilerplate, menus, scripts, advertisements, and unrelated page text. The service supports more than 200 Schema.org entity types, including Product, Article, NewsArticle, BlogPosting, LocalBusiness, Organization, Person, Event, JobPosting, Recipe, FAQPage, HowTo, Course, Book, Movie, Dataset, SoftwareApplication, Review, Service, WebPage, and many others. Full extraction is designed for pages whose existing structured data is absent, incomplete, generic, or unsuitable for the agent's task. It identifies the most useful page-level entity, preserves relevant published facts, fills a normalized Schema.org response from available page evidence, validates the resulting structure, and reports source and confidence information so callers can decide how to use the result. Common uses include product research, entity ingestion, content classification, lead enrichment, knowledge graph preparation, comparison workflows, citation collection, monitoring, and reducing tokens consumed by raw HTML. The response is compact JSON suitable for direct programmatic use. It does not return an unrestricted copy of the page and does not claim independent truth beyond the requested page. Call the free /check endpoint first for a reachability and difficulty assessment. Price: $0.01 USDC per successful request on Base through x402. Payment authorization settles only after a successful response; rejected or failed requests do not settle. API-key billing is also available for funded JSON Recon accounts.

JSON Recon Lite Schema Retrieval returns the Schema.org structured data already published on a public webpage. It is the lowest-cost JSON Recon option for AI agents, autonomous workflows, RAG ingestion, MCP clients, search tools, data pipelines, and developers that need clean JSON-LD without processing an entire page. Submit a URL and receive the best useful published Schema.org entity in a concise machine-readable response, avoiding navigation, boilerplate, menus, advertisements, scripts, and unrelated page content. Lite is ideal when a page already provides usable structured data and the caller wants speed, predictable output, and minimal cost. It can retrieve published entities such as Product, Article, NewsArticle, BlogPosting, LocalBusiness, Organization, Person, Event, JobPosting, Recipe, FAQPage, HowTo, Course, Book, Movie, Dataset, SoftwareApplication, Review, Service, WebPage, and many other Schema.org types. Page-level entities are preferred over generic site-level records when both are present. Foreign vocabulary types are not presented as Schema.org primary entities. This endpoint preserves existing published fields rather than generating replacement facts. That makes it useful for catalog discovery, content classification, entity ingestion, metadata normalization, knowledge graph preparation, product research, monitoring, and agent tool calls where native JSON-LD is sufficient. If a page has no useful published Schema.org data, use Full Schema Extraction instead. The output reflects the requested page and is not an independent guarantee of accuracy. Call the free /check endpoint before paying for a reachability and difficulty assessment. Price: $0.005 USDC per successful request on Base through x402. Payment authorization settles only after a successful response; rejected or failed requests do not settle. API-key billing is also available for funded JSON Recon accounts.

JSON Recon Verified Product checks Product facts against multiple representations on the requested public webpage and returns a concise, evidence-backed Schema.org Product record for AI agents, catalog tools, comparison systems, RAG pipelines, MCP clients, and application developers. It is designed for decisions where extracting a Product object is not enough and the caller needs to know whether important fields agree across the page at observation time. Each successful response includes the normalized Product, a verdict of verified, conflicted, or insufficient_evidence, field-level corroboration status, bounded evidence excerpts, detected contradictions, observation time, content SHA-256, coverage counts, and an estimate of context tokens avoided. Evaluated fields can include product name, brand, price, price currency, availability, SKU, MPN, GTIN, and canonical product URL when those facts are present. Sensitive URL credentials, query strings, and fragments are excluded from the public result. Use Verified Product to detect price or availability mismatches, compare visible claims with published structured data, assess catalog records before ingestion, support shopping and sourcing agents, reduce hallucination risk from loosely parsed pages, and replace large webpage payloads with compact machine-readable evidence. A conflicted result is intentionally returned rather than hiding disagreement. Insufficient evidence indicates that the page qualified as a Product but did not provide enough independent support for a stronger verdict. Verification means internal page consistency at the recorded observation time. It is not an independent guarantee of merchant truth or authenticity. Non-product pages return HTTP 422 and do not settle. Call the free /check endpoint first for reachability. Price: $0.02 USDC per successful result on Base through x402. Payment authorization settles only after a successful response; rejected or failed requests do not settle.

Return the Schema.org structured data already published on a public web page. Navigation-only results are marked low information so agents can choose Full extraction. $0.005 USDC per successful request on Base. Call /check before paying. Payment authorization settles only after a successful response.