Agent & Infrastructure

Live funding hedge desk: rates, carry side hints, hedge candidates for agent routers.

Buyable: one URL, one card. Unpaid 402 dry-read. Says if a stock x402 v2 EVM client can pay. Does not settle. Does not send funds. Informational only. Not an endorsement of the resource. Not a guarantee a later settle will succeed.

Read-only JSON-RPC access to the Hyperliquid HyperEVM, served by the Pocket Network decentralised supplier set. Pay per request in USDC; no account, no API key.

Score up to 10 x402 endpoints in a single paid call. Each found result carries the FULL trust report — identical to POST /v1/x402-trust: score, grade, scoreRange, recommendation, confidence breakdown, the component/subscore breakdown, explanations, flags + flagsDetailed, advertised price, and 30-day on-chain stats — not just the scalar verdict. Cache rows older than 15 minutes are recomputed on-demand from the latest stored probes and settlements, so scores typically reflect reality within minutes. No live network re-probe is performed — for a fresh live probe use /v1/x402-trust. Per-request recompute limits apply (time + count); results include `recomputed` and `computed_at` so you can see which rows used cached vs freshly computed scores. Resources not in our observation set return `found:false`; you still pay for the batch AS LONG AS at least one resource is tracked. If NONE of the supplied resources are in our observation set the request is rejected with HTTP 400 `notCharged:true` and you are NOT charged. Duplicates are ignored.

Is this dependency safe to use? Pass a package name (npm, PyPI, Go, Maven, crates.io, RubyGems, NuGet) and optional version: returns known vulnerabilities (OSV/CVE with CVSS and fixed-in version), deprecation status, license, latest version, repo health (stars, OpenSSF Scorecard) and an overall ok/caution/avoid verdict. JSON response. $0.01 USDC per call, pay via x402, no API key.

Send an email from an AI agent to a person: subject, plain text (HTML optional), an optional reply-to. Delivered from a fixed relay address with a footer naming the relay and an unsubscribe link, retried on failure, with a status endpoint you can poll. Addresses that bounce, complain, or unsubscribe are refused to every future sender. Up to 4 per hour and 12 per day to one address. One-time $0.01, no account needed.

Picture summary API: describes the contents of any publicly hosted image in plain sentences, what's in it, where it's set, what's happening, and any text visible in the frame. Nothing is saved after the request completes. For content moderation queues, media library tagging, and giving non-visual agents a way to reason about images.

Run a chat completion against an open-weight 20B-class model — plain prompt in, text out, with automatic failover across independent inference backends so a single call still answers when any one provider is down or rate-limiting

Black-box check: does the x402 endpoint verify payment (reject forged txHash/signature/replay) before delivering? Buyer-side trust check before you pay. Returns SAFE/RISK/NOT_X402.

AI white-box audit of your x402 payment integration source (JS/Py/Go) against the x402 Payment Security Checklist v2 — payment verification, replay, delivery races, allowance, custody, denial. status=needs-confirmation findings.

Generate a blog post outline with an AI.

Composite transformation: extracts structured data, entities, tables, and stats from text. One call replaces 4 endpoints.

Cheapest venue to buy and richest to sell ONE pair on Polygon PoS across QuickSwap, SushiSwap and Uniswap v3. Returns buyAt and sellAt (venue, price, tvlUsd), midSpreadBps, grossSpreadBps, optimalInput, venuesConsidered. Requires ?pair=SYM/SYM (e.g. WMATIC/USDC). Errors: 400 bad_pair, 404 not_routable, 502 upstream_read_failed. Ranks venues for one pair; to rank EVERY pair use /scan. Execution at a size is /slippage. Chain-pinned; bare /route = BSC.

Advisory preflight on one endpoint before a consequential call: a bounded live probe (reachable, status, latency, MCP initialize, current x402 price and payTo) joined to our observed history (reliability over a stated window, payTo and price changes, first seen). Returns ALLOW, WARN or UNKNOWN. Never a guarantee. No evidence, no charge.

Research one x402 product idea against live 402index opportunity gaps, 402index competitors, and Coinbase Bazaar discovery, then return saturation, pricing, competitors, and a build recommendation.

Run a chat completion against an open-weight 20B-class model — plain prompt in, text out, with automatic failover across independent inference backends so a single call still answers when any one provider is down or rate-limiting

What is inflation, GDP growth or unemployment in these countries? Resolved to each country's latest published year and returned with ISO code, year, value and unit — one call for a whole country set, with no series ids to look up and no per-country round trips.

Given any public HTTP(S) URL, returns a measured liveness verdict - LIVE, PAUSED, GATED, CHALLENGED, DEGRADED, DEAD or UNKNOWN - with every reason carrying the source it was read from. The point is the case a status check misses: a front door answering 200 over a back end that stopped. Also sold at this origin: good ($149.00, POST /x402/good) - The x402 rail, ninety days - framed print, 24 x 36 in; census ($100.00, GET /x402/census) - The x402 rail census, ninety days, whole.

Generate a concise briefing on a topic using AI.

TTS API paid per call over x402: up to 2000 characters to natural speech (WAV) on our locally-hosted neural speech engine, multiple voices. No keys, no subscription. For voice agents, IVR, audio content. Premium tiers with 20 voices and 31 languages at voice.forgemesh.io.

Image description API: detailed natural-language description of any image URL — subjects, setting, actions, visible text — from our locally-hosted vision engine. Use for image understanding, content cataloging, moderation pre-screening, and vision-blind agent workflows. Images processed in memory, never stored.

Text-to-speech API: convert up to 2000 characters into natural speech (WAV, base64) using our locally-hosted neural speech engine — no API keys, pay per call with USDC. Choose voices F1-F5/M1-M5. Use for voice agents, notifications, audio content, and IVR. Text never stored. Full 20-voice / 31-language tiered product at voice.forgemesh.io.

Image captioning API: one-call natural-language caption for any image URL, generated by our locally-hosted vision engine. Ideal for cataloging, moderation triage, and dataset labeling. No data retained.

Fetch a web page and return its title and raw HTML (up to 200,000 characters). No browser engine is used and JavaScript is not executed, despite the name. POST /v1/render with {url} and get one JSON object back. Pages are cached for 5 minutes. Pay per request in USDC; no account, no API key.