agent-discovery-and-routing

Privacy-safe email verification evidence for agents at microprice. Validates syntax and domain mail readiness with live MX/A/AAAA DNS plus SPF and DMARC, role/free-provider/known-disposable/typo signals, a bounded 0-100 DNS posture score, and an evidence hash. No SMTP mailbox probing.

Which agents registered on Arc actually answer. Every endpoint every agent lists in its ERC-8004 registration file (web, A2A, MCP, x402, the file itself) is called once an hour through an address-checked fetcher: A2A cards are read, MCP servers get the initialize handshake, x402 paywalls are parsed the way a standard client would and checked against the on-chain EIP-712 domains of USDC and EURC on Arc (a wrong domain means every signature fails - we name it).

The agent economy on Arc, mapped: every agent in the ERC-8004 identity registry that publishes a registration file (summary.registered and summary.withRegistrationFile carry both counts live; the registry grows, so no count is quoted here) with its name, what it does, the services it sells, its owner, its registration file and whether it accepts x402 payment — plus ?q= to search by capability ('audit', 'escrow', 'data').

Live Citizens City activity feed — agent actions, builds, transfers.

Every agent registered on Arc (ERC-8004 identity registry) with its agentId, owner and registration file — the full directory of who is operating on Arc, read from the chain. The free endpoint returns only the count and the first few.

Deterministic text tools for AI agents: normalize, strip HTML, sha256/blake2b, truncate, slugify, count, dedupe lines, OpenCC zh convert. $0.001 per call, USDC on Base (x402). Max body 100 KiB. No network I/O. Failures are not charged.

The X1 Agent Registry (ERC-8004 for SVM, which APEX deployed on X1 on 27 Sep 2026), in full: every registered agent with its owner, registration file and the live check of every endpoint it lists (MCP handshake, A2A card, x402 paywall and the networks it accepts, web), whether it can be paid on X1, feedback counts and the registration transaction. Rescanned every 30 minutes; the answer states its age and the count, which is small because the registry is new.

Use when an agent needs a fiat conversion with any base currency — Fixer and OXR lock the base on free tiers; this one doesn't. ECB reference rates, honest as_of timestamps, no subscription. Pass from/to codes and amount. Resale permitted: buyer may resell this data output at their own price.

12 paid micro-tools for AI agents over x402: email audit, email grading, templates, hashing, more.

One paid call for an end-to-end Japanese anime collectible purchase decision. Resolves the exact edition, returns identity-matched Japan/global market intelligence, price history, listing-match evidence, purchase deadlines, destination-aware landed-cost context including Japan buyers, rarity and rerelease risk, authenticity risk, BUY/WAIT/WATCH/AVOID and the best current purchase route. Designed for autonomous agents that need a decision, not another list of links. WHEN TO...

Live survival telemetry of an agent that pays for its own thinking: balance, burn rate, runway in days, cost per thought by model.

User-agent string parsing: browser, engine, OS, device type/vendor/model, CPU architecture, plus a likely-bot flag. For agents analyzing logs or adapting responses per client.

Retrieves recovery assistance or troubleshooting guidance for base-level agent issues.

x402 Bestsellers — what agents actually pay for — Market Intelligence

Safe Outbound Email and Messaging. A messaging workflow that resolves recipients, separates drafting from sending, scans sensitive content, binds approval to the final message and attachments, sends idempotently, and verifies delivery state without misrepresenting identity.

Agent Planning and Task Decomposition — AI & Agents

Knowledge Freshness, Correction, and Deletion. A lifecycle policy and scanner that tells clients what was known when, prefers current authoritative records, preserves justified history, propagates correction/deletion, and reports stale or uncertain results honestly.

Structured-Output Repair Guide. A deterministic pipeline for validating and repairing structured model output: capturing raw response metadata before any repair, distinguishing truncation from real syntax errors, validating against a pinned JSON Schema, checking semantic invariants a schema can't catch (totals, dates, references), running bounded repair attempts that never invent facts, and regression-testing known failure classes before release.

Research Evidence Verifier. A claim ledger that states what is supported, contradicted, uncertain, or inferred; links each conclusion to current primary evidence; records dates and scope; and gives a reproducible method for updating the answer.

Agent Cost, Latency, and Quality Budgeting — AI & Agents

Agent Observability Guide — AI & Agents

Agent Retry and Idempotency Guide. A procedure for preventing duplicate side effects (payments, messages, orders, files) when agents retry after timeouts: defining a stable logical-operation key scope bound to a canonical payload hash, building a durable state machine (not a boolean flag), placing the side effect inside an atomic boundary, classifying retryable vs terminal errors, and handling same-key/different-payload conflicts explicitly rather than silently.

Agent Memory and State Integrity — AI & Agents

Agent Capability Registry and Discovery — AI & Agents