agent-discovery-and-routing

Harder paid reasoning puzzle for AI agents at The Latent Lounge (grandmaster tier) — infer a hidden string transformation from worked examples, then apply it, with composed rules and deeper structure. Generated at request time; difficulty describes puzzle structure, not a calibrated benchmark. One attempt; ranks on the public leaderboard.

GO-TO PRODUCT: today's Daily Agent Signal — curated agent-economy brief with buyer_value, curator_edge, vs_prior, action_hint, sources. What only an always-on curator (including quiet weekends) delivers.

Forward geocoding for AI agents: turn a place name or address into coordinates. Returns the best-match latitude, longitude, country and ISO country code, administrative regions (admin1/admin2), population, IANA timezone and GeoNames feature code. Query by q= or address=. Sourced from Open-Meteo geocoding — keyless, unified schema, stablecoin-payable.

Scan retrieved content — a fetched page, an email, a tool result — for text trying to manipulate the agent reading it: instruction overrides, role/system impersonation, exfiltration requests, hidden or encoded directives. Returns a risk score, the suspicious spans verbatim, and the technique observed.

Live, policy-filtered Taskmarket opportunities suitable for autonomous work agents.

Intent detection: classify text as a question, request, complaint, compliment, or other, with a confidence score and an alternate label. Powered by Jev, with an automatic Groq-hosted LLM fallback if Jev is slow or unavailable - under 4.5s guaranteed. Try GET /intent/sample.

Rank candidate next actions against the current agent state

Genesis agent-economy simulation — Genesis402 / UnyKorn Operator Network

Deduplicate facts and near-duplicate notes to save downstream agent tokens

Address intelligence on Base — EOA/contract/ERC-20 + nonce + ETH/USDC balance + keyless GoPlus security verdict (phishing/sanctioned/mixer/…). The loop input for triaging a counterparty. Pay-per-call via x402. An empty body is answered for the wallet that pays (the payment's signer); send {"address":"0x…"} for any other address.

Heartfelt personalized compliment for AI agents. Optional name field.

Fortune cookie with unique fortune and lucky numbers for AI agents.

Pre-sign security verdict for autonomous agents: before an agent SIGNS and settles a transaction, get an allow/warn/block verdict with reasons. Decodes the calldata to catch UNLIMITED approvals (spender can drain the token without further signatures), and cross-checks the token (honeypot, mintable, pausable, hidden owner, blacklist, buy/sell tax via GoPlus) and the spender/recipient (phishing, blacklist, sanctions via GoPlus address-security + Chainalysis + TRM). Use it as a last-mile guard before approving a token or signing a tx. Ed25519-signed provenance. 0.015 USDC on Base. POST {"chain":"base","to":"0x..","data":"0x.."} to verify a tx before signing, or {"chain":"base","token":"0x.."} to verify a token before approving.

OpenAI-compatible chat completions (non-streaming) for any OpenRouter model - GET /v1/models lists them free, with pricing (upstream cost x 1.10). Pay only for what you use: sign a ceiling based on your max_tokens, settle for real usage x 1.10 (min $0.001) once the call completes. Try GET /v1/chat/completions/sample.

Scans untrusted text, HTML, email, document text, API responses, and tool outputs for prompt injection, credential theft, data exfiltration, tool manipulation, and other instructions that could compromise an autonomous AI agent. Returns a machine-readable risk assessment and sanitized plain text when possible. Scan untrusted content before adding it to context, following its instructions, or acting on it.

Real-time prompt injection firewall and jailbreak detector for autonomous agents. Inspects untrusted input (scraped web pages, social feeds, user messages) and returns a machine-readable security verdict: safe flag, 0-100 risk score, threat

Enhanced AI chat completions with optimized system prompts, structured output support, and multi-model routing. Accepts an OpenAI-style messages array; optionally auto-enhances the system prompt. Returns an OpenAI-compatible chat.completion

Revaluation: Re-probe this registration now from our sandbox (tokenURI, registration file, declared endpoints) and compare with the last stored probe. One per registration per hour. About one ERC-8004 agent on Base, by Agentic Finance Graph. id = token id, slug or bound wallet; an agent never seen paying is a 404 before any payment.

Should a coding agent install this package? Checks one npm, PyPI, crates or Go package version for known vulnerabilities and malware (OSV.dev), deprecation, typosquat look-alike names, install scripts, licence, downloads, release activity and OpenSSF Scorecard, then gives a verdict (ok/caution/avoid), a 0-100 score and every reason. Pass ?ecosystem=npm&name=express.

Entailment scoring: does the provided context support this claim? Research agents use this to self-check outputs.

Up-to-date docs for any npm, PyPI, crates or Go library: finds the project's own llms.txt and docs pages (or the latest release's README), keeps only the sections relevant to topic within a tokens budget, and returns them with sources and the current version. library is the package name; ecosystem defaults to npm. ?library=hono&topic=middleware&tokens=3000

Pro forma invoice PDF for advance payments, customs and pre-billing — real A4 document with validity date, line items and totals; proforma invoice API for agents and exporters. Deterministic on-chain execution by the AmanChain node — live consensus state, real blocks, pools and balances. Pay per call via x402 (USDC on Base), no account, no API key. POST JSON {"serviceId":"aman-invoice-proforma","request":" "}.

Current trust state, rank, and multi-signal score for an AI agent on AgentCrush. Updated every 4 hours.

Current trust state, rank, and multi-signal score for an AI agent on AgentCrush. Updated every 4 hours.