{"name":"AgentIndex content kit","description":"Process the files and content an agent gives you. Nothing else.","routes":[{"route":"/detect-language","method":"GET","price":"free","summary":"Detect the language of a piece of text - the kit's free entry point.","use_cases":["check that the kit is reachable before paying for anything else"],"sample":null,"input_example":null,"output_example":null},{"route":"/discover","method":"GET","price":"free","summary":"Find MCP servers by need, ranked by semantic relevance over a curated snapshot — no payment (5 by default, max 10 via max_results).","use_cases":["discover MCP servers matching a need before wiring a client","free snapshot search when POST /discover payment is not needed"],"sample":"/discover/sample","input_example":{"q":"postgresql jdbc read only mcp"},"output_example":{"q":"read a PDF and give me markdown","snapshot_date":"2026-09-18","snapshot_rows":10101,"min_similarity":0.3,"matches":3,"results":[{"name":"PDF Extract MCP","url":"https://example.org/mcp/","description":"Extract a public PDF into clean markdown text, plus metadata and a real token count.","registry":"registre-mcp","relevance":0.7211,"observed_at":"2026-09-02T19:07:49.374328Z"},{"name":"Docling Server","url":null,"description":"Document conversion to markdown and structured text for LLM agents.","registry":"annuaire","relevance":0.6487,"observed_at":"2026-09-10T12:00:00Z"},{"name":"Srclight","url":"https://example.com/mcp/","description":"Deep code indexing for AI agents. FTS5 + embeddings + call graphs. Fully local.","registry":"registre-mcp","relevance":0.4102,"observed_at":"2026-09-15T08:30:00Z"}]}},{"route":"/discover","method":"POST","price":"$0.001","summary":"Match a need in plain language against a curated snapshot of 10101 observed MCP servers and get the best five with a 0-1 relevance score each - semantic ranking a web search cannot guarantee.","use_cases":["find the MCP server that fits a need instead of guessing its name","pick from the best-matching servers with a relevance score before contacting one","discover servers for a job a web search ranks by keywords, not meaning","check what kinds of MCP servers exist for a capability before building one"],"sample":"/discover/sample","input_example":{"q":"read a PDF and give me markdown"},"output_example":{"q":"read a PDF and give me markdown","snapshot_date":"2026-09-18","snapshot_rows":10101,"min_similarity":0.3,"matches":3,"results":[{"name":"PDF Extract MCP","url":"https://example.org/mcp/","description":"Extract a public PDF into clean markdown text, plus metadata and a real token count.","registry":"registre-mcp","relevance":0.7211,"observed_at":"2026-09-02T19:07:49.374328Z"},{"name":"Docling Server","url":null,"description":"Document conversion to markdown and structured text for LLM agents.","registry":"annuaire","relevance":0.6487,"observed_at":"2026-09-10T12:00:00Z"},{"name":"Srclight","url":"https://example.com/mcp/","description":"Deep code indexing for AI agents. FTS5 + embeddings + call graphs. Fully local.","registry":"registre-mcp","relevance":0.4102,"observed_at":"2026-09-15T08:30:00Z"}]}},{"route":"/extract","method":"POST","price":"$0.05","summary":"Extract structured data from a URL or text into strict JSON matching a schema you provide - never an approximate or invented value.","use_cases":["pull structured fields (price, name, date, etc.) out of a product or listing page","convert a page or block of text into JSON matching your own schema","extract contact information or structured facts from a document","parse a web page into the exact fields your application needs, nothing else","get an explicit failure instead of a guessed value when a field isn't present","extract structured data from raw text without writing your own parser","turn an unstructured page into a strictly-typed JSON record","validate that extracted data actually matches your schema before using it"],"sample":"/extract/sample","input_example":{"url":"https://example.com/product/widget-pro","schema":{"type":"object","properties":{"product_name":{"type":"string"},"price":{"type":"number"},"in_stock":{"type":"boolean"}}}},"output_example":{"data":{"product_name":"Widget Pro","price":29.99,"in_stock":true}}},{"route":"/fact-check","method":"POST","price":"$0.05","summary":"Check a claim against live web sources and get a verdict, a confidence level, and the supporting or contradicting sources.","use_cases":["verify a claim against live web sources before including it in an answer","get a confidence level alongside a verdict, not just a yes or no","find sources that support or contradict a specific factual claim","check whether a number or statistic is still accurate today","fact-check a claim from a document, chat, or user-submitted text","gather citable URLs for or against a claim before writing a report","distinguish a well-supported claim from an unverifiable one","check a claim before repeating it in a high-stakes context"],"sample":"/fact-check/sample","input_example":{"claim":"The Eiffel Tower is taller than the Statue of Liberty."},"output_example":{"claim":"The Eiffel Tower is taller than the Statue of Liberty.","verdict":"supported","confidence":0.92,"sources":[{"url":"https://example.com/eiffel-tower-facts","title":"Eiffel Tower Facts","stance":"supports"}]}},{"route":"/jobs","method":"POST","price":"$1.00","summary":"Delegate a multi-step research task and get back one cited, sourced brief instead of running many searches yourself.","use_cases":["delegate a multi-step research task and get back a single cited brief","get a sourced synthesis of a topic instead of running many searches manually","research a company, person, or topic and receive a structured summary with sources","offload long-running research so it keeps running in the background","produce a multi-source report on a subject without reading every page yourself","poll a research job for status and retrieve the result once it is ready","build a background brief on a subject before a meeting or decision","compile findings from many sources into one answer with citations"],"sample":"/jobs/sample","input_example":{"subject":"Example Corp"},"output_example":{"job_id":"abc123","eta_seconds":60}},{"route":"/pdf","method":"POST","price":"$0.05","summary":"Extract a PDF (by URL or base64) into clean markdown with metadata and a real token count - deterministic, no model.","use_cases":["extract clean text from a PDF report or paper without running your own PDF parser","pull the title, author and page count from a PDF before deciding whether to read it fully","convert a PDF invoice or contract into markdown for further processing","get an accurate token count for a PDF before feeding it into a model's context window","read a PDF linked from a web page by URL, no download step of your own","process a base64-encoded PDF received from another tool or upload","extract text from academic papers or whitepapers for summarization or search","check a PDF's page count and metadata before committing to a full read"],"sample":"/pdf/sample","input_example":{"url":"https://example.com/report.pdf"},"output_example":{"markdown":"Universal Declaration of Human Rights \nPreamble \nWhereas recognition of the inherent dignity and of the equal and inalienable \nrights of all members of the human family is the foundation of freedom, justice \nand peace in the world, ...","metadata":{"title":null,"author":"lindner","pages":8,"date":"2001-09-10T01:04:58-17:00"},"token_count":2330}},{"route":"/search","method":"POST","price":"$0.01","summary":"Real-time web search with ranked results, a cleaned page extract, and an optional short answer - up to 5 queries merged and de-duplicated in one call.","use_cases":["search the live web for recent information on a topic","verify a claim against current sources before including it in an answer","run several related queries in one call and get a single merged, de-duplicated result set","pull the latest news or announcements about a company, person, or event","gather source URLs and short extracts to cite in a report","check whether a fact or number has changed since a model's training cutoff","compare how different sources describe the same topic without making separate calls","collect background material on a subject before writing a summary"],"sample":"/search/sample","input_example":{"query":"best ramen restaurants in Shibuya Tokyo"},"output_example":{"query":"best ramen restaurants in Shibuya Tokyo","results":[{"title":"The Best Ramen in Tokyo: 16 Restaurants to Not Miss | Eater","url":"https://www.eater.com/maps/best-ramen-tokyo-japan","date":null,"extract":"A curated guide to Tokyo's standout ramen shops, including several in Shibuya..."}]}},{"route":"/summarize","method":"POST","price":"$0.03","summary":"Summarize a URL, text or HTML at the length you request, neutral and factual, with the source noted.","use_cases":["get a short summary of a long article before deciding whether to read it fully","summarize a URL, raw text or HTML at the exact length you need","produce a one-paragraph summary of a document for a report or digest","summarize search results or scraped content without writing your own prompt","condense a long page into a few sentences while keeping it factual","get a source-noted summary suitable for citing in a longer answer","summarize HTML content directly without fetching or cleaning it yourself","produce consistent-length summaries across many documents in a pipeline"],"sample":"/summarize/sample","input_example":{"url":"https://example.com/photosynthesis","length":"short"},"output_example":{"summary":"The article explains how photosynthesis converts light energy into chemical energy in plants.","length":"short","sources":["https://example.com/photosynthesis"]}},{"route":"/translate","method":"POST","price":"$0.10","summary":"Translate one string or a batch of up to 200 text segments in a single call, preserving markdown, HTML and {placeholder} formatting.","use_cases":["translate a whole file or document in one call instead of one request per line","localize UI strings or interface text into a target language","translate a batch of text segments while keeping their order and count intact","preserve markdown formatting and {placeholder} tokens when translating templated text","detect the source language of unlabeled text before translating it","convert user-submitted content into another language for moderation or display","translate a list of product descriptions or messages in bulk","get a translated version of text alongside the original for comparison"],"sample":"/translate/sample","input_example":{"text":"Bonjour le monde","target_lang":"en"},"output_example":{"text":"Bonjour le monde","target_lang":"en","detected_source_lang":"fr","translated_text":"Hello world","model_served":"agentindex-translate-1"}},{"route":"/web-read","method":"POST","price":"$0.03","summary":"Fetch a URL and return its main content as clean markdown, boilerplate stripped, with a real token count.","use_cases":["read a specific article or page by URL and get clean markdown back","strip navigation, ads and boilerplate from a page before feeding it to a model","resolve a page's main content when you already have the URL, without a search step","get an accurate token count for a page before including it in a model's context","pull the readable content of a news article, blog post or documentation page","extract a page's title alongside its cleaned body content","read a source page found by another tool or by a human-supplied link","convert an arbitrary web page into markdown for storage or further processing"],"sample":"/web-read/sample","input_example":{"url":"https://www.w3.org/"},"output_example":{"url":"https://www.w3.org/","title":"World Wide Web Consortium (W3C)","markdown":"# World Wide Web Consortium (W3C)\n\nThe W3C mission is to lead the web to its full potential.","token_count":154}}]}