Give an AI agent structured answers instead of raw pages
For: teams building AI agents and assistants that must answer from live facts. One run is 4 calls and 10 credits.
Get an API key Run a step free
The job
An agent that reads whole web pages spends its context on navigation, ads and layout, and still has to guess which number is the one you asked about.
Web data or web intelligence?
Web data is the page; web intelligence is the answer. Each call returns named fields, so a tool result is short, predictable and ready to quote: a repository's description, a paper's title and abstract, a coin's price, a company's status. You define one tool per endpoint and the agent reads fields, not pages.
The calls, in order
| Step | Call | What it gives you | Credits | Docs |
|---|---|---|---|---|
| 1 | POST /v1/developer/github/repo | tool one: what is this repository? | 1 | Call and code |
| 2 | POST /v1/science/pubmed/search | tool two: what has been published on this? | 5 | Call and code |
| 3 | POST /v1/crypto/coingecko/coin | tool three: what does this coin cost now? | 2 | Call and code |
| 4 | GET /v1/companies/search | tool four: is this company registered, and what type is it? | 2 | Call and code |
What one run costs
fact The credits per call are the catalog's: 1 + 5 + 2 + 2 = 10 per run, so 10,000 credits per 1,000 runs. calculated At the plan prices on the pricing page that is about $79.60 per 1,000 runs on Production and $98.00 on Builder (plan price divided by its monthly credits, times the credits). The Free plan gives 100 credits a month with no card.Formula
Your runs a month
Real responses, step by step
fact Captured from https://api.foxapis.com on 19 November 2026 with the documented sample input of each call; every value below is from that response, shortened where long. The nightly check on the status page lists each of these calls as working.
Step 1: POST /v1/developer/github/repo
The input: {"url":"https://github.com/torvalds/linux"}. A real response had these fields.
| Field | Value in that response |
|---|---|
title | torvalds/linux |
author_name | torvalds |
published_at | 2011-09-04T22:48:12Z |
modified_at | 2026-11-18T20:23:07Z |
likes | 256,193 |
views | 256,193 |
curl -X POST 'https://api.foxapis.com/v1/developer/github/repo' \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"url":"https://github.com/torvalds/linux"}'
Step 2: POST /v1/science/pubmed/search
The input: {"query":"crispr","max_results":3}. A real response had these fields.
| Field | Value in that response |
|---|---|
total_results | 72,829 |
returned | 3 |
curl -X POST 'https://api.foxapis.com/v1/science/pubmed/search' \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"query":"crispr","max_results":3}'
Step 3: POST /v1/crypto/coingecko/coin
The input: {"coin_id":"bitcoin"}. A real response had these fields.
| Field | Value in that response |
|---|---|
coin_id | bitcoin |
name | Bitcoin |
symbol | BTC |
rank | 1 |
price_usd | 62,529 |
price_change_24h_pct | -2.278 |
curl -X POST 'https://api.foxapis.com/v1/crypto/coingecko/coin' \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"coin_id":"bitcoin"}'
Step 4: GET /v1/companies/search
The input: {"q":"stripe","page_size":3}. A real response had these fields.
| Field | Value in that response |
|---|---|
company_number | 16501035 |
company_name | STRIPE & BLOOM LTD |
company_status | active |
company_type | Private Limited Company |
incorporation_date | 2025-06-06 |
jurisdiction | united kingdom |
curl 'https://api.foxapis.com/v1/companies/search?q=stripe&page_size=3' \
-H "Authorization: Bearer YOUR_API_KEY"
All of it as one script
The same 4 calls chained, with a key placeholder. Swap the inputs for your own.
// 4 calls, one job run. Put your key in YOUR_API_KEY.
const KEY = "YOUR_API_KEY", BASE = "https://api.foxapis.com";
async function call(method, path, input) {
const url = BASE + path + (method === "GET" ? "?" + new URLSearchParams(input) : "");
const r = await fetch(url, { method, headers: { Authorization: "Bearer " + KEY, "Content-Type": "application/json" }, body: method === "GET" ? undefined : JSON.stringify(input) });
if (!r.ok) throw new Error(path + " answered " + r.status);
return r.json();
}
const step1 = await call("POST", "/v1/developer/github/repo", {"url":"https://github.com/torvalds/linux"}); // tool one: what is this repository?
const step2 = await call("POST", "/v1/science/pubmed/search", {"query":"crispr","max_results":3}); // tool two: what has been published on this?
const step3 = await call("POST", "/v1/crypto/coingecko/coin", {"coin_id":"bitcoin"}); // tool three: what does this coin cost now?
const step4 = await call("GET", "/v1/companies/search", {"q":"stripe","page_size":3}); // tool four: is this company registered, and what type is it?
console.log({ step1, step2, step3, step4 });
What you could build
- One tool per endpoint in your agent framework, with the input from the sample below as its schema
- A rule that the agent quotes the returned fields and says when a field is empty
- A cost guard: credits per call are known, so you can cap an agent run
Limits, plainly
These four tools are a starting set from endpoints that worked in our nightly check. The company search covers United Kingdom registered companies. Each call is independent; keep your key on the server that runs the agent.
Field lists and the rest of each data type: Developer fields · Research fields. Single calls with code in six languages: task APIs.
Other jobs: Track a product's open-source and launch momentum · Build a crypto and DeFi dashboard · Write a biomedical research brief · Analyse players and games for a gaming community · Watch artists for a label or a playlist curator · Enrich a lead list with company and developer signals · Check a UK company and who controls it · Compare your app with a rival's on both phone stores · Profile a creator before a sponsorship · Check an open-source model and its repository before you adopt it · Profile a developer expert before you hire or sponsor · Check a person's claims before you hire or partner · Score a candidate against a role you describe · Prepare a sales brief on a company and score a prospect's message · Check how AI assistants show your domain · Research an investor before you pitch.
Source: live calls to https://api.foxapis.com on 19 November 2026. Values reflect that moment and those inputs. Credits per call are the catalog's.
How we calculate
- Cost per run and per 1,000 runs: The credits of each call in the run, added up (the catalog's credits per call), times 1,000 for 1,000 runs.
- Dollar figures: The plan's monthly price divided by its monthly credits, times the credits. Prices come from the plans file at build time and are never typed.
Data: the FoxAPIs catalog and plans, read 19 November 2026. Every other figure on this page is copied from the record, not calculated by us.
Run this free, then pay per call
Free to start: 100 credits, no card. Then from $10. Credits you buy never expire. Use them at your own pace. Get my free key
One call costs 1 credit. That is $7.96 per 1,000 calls on the Production plan and $9.80 on Builder (plan price divided by its monthly credits). The Free plan gives 100 credits a month with no card.
Building an AI agent?
- The FoxAPIs MCP server gives your agent every tool through one connection.
- Per-call pricing in credits: you pay for the calls your agent makes, with no seats.
- Free plan: 100 credits a month, no card, no monthly fee.