canonical: https://jentic.com/apis/neurobird.com/neurobird # Neurobird Search API The Neurobird Search API searches the live web for AI agents in a single call: it fans a query out across many search engines, fuses the rankings with reciprocal rank fusion, fetches and extracts the top pages, and ranks the passages inside them. It can optionally write an answer and, with verify_quotes, check each supporting quote against its source text. A separate extract operation turns one or more URLs, including PDFs, into clean markdown or text, and the request and response shapes match Tavily so an existing Tavily client switches by changing its base URL and key. ## For AI agents Search the live web with ranked, pre-extracted passages and optional quote-checked answers, stream each retrieval stage, and extract clean markdown from URLs. ## Scope Does not crawl entire sites, host a private document index, or generate embeddings. Use for live web search and page extraction only. ## Capabilities - Search the live web and get ranked results with the relevant passages already extracted, at a search_depth of basic, standard, or advanced - Narrow a search by topic such as news, code, science, or finance, and by include_domains, exclude_domains, country, language, or time_range - Generate a grounded answer with include_answer and have verify_quotes check every supporting quote against its source - Stream each stage of the retrieval pipeline as server-sent events while a search runs - Extract the main content of one or more URLs as clean markdown or text, converting PDFs and stripping navigation and adverts - Create a free key with no signup, then read usage, credit balance, and credit pack prices for that key ## Use cases ### Research agent with grounded web answers A research agent answering a factual question needs current sources, not model memory. Through Jentic the agent searches for 'search the web', loads the Neurobird search operation with its input schema, and posts a query with include_answer and verify_quotes enabled. The response carries ranked results with their relevant passages already extracted, so the agent usually skips a follow-up fetch, plus an answer whose quotes were checked against the source text. The Neurobird key is attached by your own Jentic One instance at call time. Example prompt: Call POST `/v1/search` with query 'EU AI Act enforcement timeline', topic 'news', search_depth 'advanced', include_answer true, and verify_quotes true, then report the answer with its sources ### Clean page extraction for downstream processing A summarization pipeline needs the article body of a list of links without navigation, adverts, or boilerplate. The extract operation fetches one or more URLs and returns the main content as markdown or plain text, converting PDFs to text, and it costs one credit per page. Example prompt: Call POST `/v1/extract` with urls containing two article links and format 'markdown', then summarize each result ### Drop-in replacement for a Tavily client A developer already using Tavily wants to compare results or costs. Because Neurobird's request and response shapes match Tavily, the existing client switches by pointing at https://search.neurobird.com and sending a Neurobird key. Credits are 1 for a basic or standard search and 2 for an advanced one, with a grounded answer or structured extraction adding 1. Example prompt: Send the same query to POST `/v1/search` with search_depth 'basic', then call GET `/v1/usage` to read the credits consumed ### Watching retrieval progress live An interface that shows users what the agent is doing streams the search instead of waiting for the whole result. The streaming operation emits each pipeline stage as it completes, from engine results to fetched pages and ranked passages, over server-sent events. Example prompt: Call POST `/v1/search/stream` with query 'best open source vector databases 2026' and relay each stage event to the user as it arrives ## Key endpoints | Method | Path | Description | | --- | --- | --- | | POST | `/v1/search` | Search the web | | POST | `/v1/search/stream` | Search with streamed stages | | POST | `/v1/extract` | Extract clean page content | | POST | `/v1/keys` | Create a free API key | | GET | `/v1/usage` | Usage for the calling key | | GET | `/v1/billing/balance` | Credit balance for the calling key | | GET | `/v1/billing/packages` | Credit packs and request costs | ## Key resources - **Search** — Web search with extracted passages, optional answers and quote checking, streamed stages, and a paid per-call variant - **Extract** — Clean markdown or text extraction from one or more URLs, including PDFs - **Keys and usage** — Free key creation with no signup and per-key request count, credits consumed, and average latency - **Billing** — Credit packs with prices, credit balance split into free grant and purchased credits, card and crypto checkout, and payment options - **Health** — Retrieval backend reachability, active egress addresses, and answer generation availability ## AI readiness This API is usable in Jentic One now. Its AI-readiness score against Jentic's framework shows where it stands today and where improvements would make it even easier for agents to use. - **Score:** 62 / 100 - **Maturity:** AI-Aware - **Dimensions:** - Foundational Compliance: 83 / 100 - Developer Experience & Jentic Compatibility: 57 / 100 - AI-Readiness & Agent Experience: 44 / 100 - Agent Usability: 94 / 100 - Security: 55 / 100 - AI Discoverability: 77 / 100 - **View full report:** https://jentic.com/apis/neurobird.com/neurobird/scorecard - **How the score is calculated:** https://docs.jentic.com/reference/api-readiness-framework/overview/ - **More about the dimensions:** https://docs.jentic.com/reference/api-readiness-framework/specification/#dimensional-model-overview ### Score it yourself Every API in the directory is allowlisted, so you can re-score it with no key required. - **Score your own API:** https://jentic.com/scorecard.md - **Scoring CLI agent skill:** https://github.com/jentic/jentic-api-scorecard/blob/main/skills/jentic-api-scorecard/SKILL.md ```sh npx @jentic/api-scorecard-cli score ``` ## Why Jentic - **Setup:** Wiring Neurobird by hand means minting an nbs- key, choosing between bearer and x-api-key headers, telling free search routes apart from the x402 paid route, and handling server-sent events for streamed search. Through Jentic you install once, import Neurobird from the API Directory, store the key once, and your agent calls it. - **Permission scoping:** Neurobird carries the query and URLs inside the request body, so rules bound which operations your agent may call, not what it searches for. Limit the agent to the operations it needs, such as search and extract, so spending ones like card or crypto checkout are not included unless you add them. - **Credential handling:** Your Neurobird key is stored once, encrypted, by your own Jentic One instance and injected at execution time. It never enters the agent's prompt, logs, or context. - **Discovery method:** Agents search Jentic by intent such as 'search the web' or 'extract a page as markdown', and Jentic returns the matching Neurobird operation with its input schema so the agent sets search_depth and filters correctly. ## Related APIs - **Tavily API** — Web search API for agents with the same request and response shapes - **Exa** — Neural web search and content retrieval API - **Firecrawl API** — Crawls whole sites into clean markdown ## FAQ ### What authentication does the Neurobird Search API use? Per its OpenAPI spec, Neurobird keys look like nbs- followed by the key and are sent as a bearer token in the Authorization header, or in an `x-api-key` header for clients that prefer it. Usage, credit balance, and checkout operations use the key, while the search, extract, key creation, packages, payments, and health operations are declared with no security requirement. The paid search variant can instead be paid per call with x402 or MPP. Through Jentic your key is stored encrypted by your own instance and attached per call. ### Can I get an answer with verified quotes from the Neurobird Search API? Yes. Set include_answer on POST `/v1/search` to have Neurobird write an answer from the ranked passages, and set verify_quotes to check each supporting quote back against its source text. A grounded answer adds 1 credit to the cost of the search. ### What are the rate limits for the Neurobird Search API? The spec prices requests in credits rather than stating request-rate ceilings: a free key from POST /keys carries 1,000 free credits a month, a basic or standard search costs 1 credit, an advanced search 2, and /extract 1 per page. It notes the keyless evaluation path is throttled. GET `/v1/usage` and GET `/v1/billing/balance` report what your key has used and has left. ### Is there a Neurobird MCP server? Yes - Neurobird offers a hosted MCP server at POST /mcp exposing web_search and extract_url. You can also connect the Neurobird Search API directly through Jentic: your agent gets the full 27-endpoint surface without loading another server's tool definitions into its context, your key is injected at call time by your self-hosted instance instead of living in an MCP config, and your rules decide which operations the agent may call. ### How do I search the web with Neurobird through Jentic? Import Neurobird from the API Directory in your Jentic One instance and store your nbs- key. Your agent searches Jentic for 'search the web', loads POST `/v1/search`, and sends a query with options such as topic, max_results, and include_domains. To run it on your own infrastructure, install Jentic One from its GitHub repo. ### Can I limit what my agent is allowed to do with the Neurobird Search API? Yes. Write a rule that allows only POST `/v1/search` and POST `/v1/extract`, and the agent can search and read pages but cannot start a card or crypto checkout or call the per-call paid search route. Every call it makes is logged.