Give an agent a bounded, deterministic check at the moment it needs one. Discover schemas for free, inspect the payment requirement before spending, and pay only for the execution you choose.
The gateway separates discovery from execution. Product descriptions, input schemas, prices, OpenAPI metadata, MCP tools, and deterministic sample fixtures are public. A paid route first returns exact x402 payment terms; execution follows only after the client supplies a valid payment.
Twenty-two utilities cost $0.05 per call. Seven deeper evidence packets cost $0.25, including Agent Distribution Readiness, Community Rule & Credibility Preflight, Qualified Demand Ledger, and Channel-Fit Evidence Matrix.
Shopify CSV, image-to-SKU matching, UTM links, gitignore coverage, unsubscribe headers, npm publish payloads, redirect maps, subtitle timelines, calendar time zones, filename portability, CSV schema drift, release artifact manifests, and email DNS authentication evidence.
Promise-to-action evidence, tool-call contracts, completion receipts, tool descriptions, skill activation, cost decomposition, API drift, webhook replay, and policy-change evidence.
Audit an agent run; check product readiness; rank channel tests against evidence and qualified gates; classify demand without counting vanity activity; preflight a proposed community action against rule and credibility evidence; query bounded Chicago demolition-permit matches; or compare dated site-discovery evidence.
@justhandledlabs/agent-client@0.8.0 for controlled paid execution.io.github.justhandledlabs/agent-gateway in the official MCP Registry.Twenty-nine deterministic preflights, maintained-data lookups, and evidence packets across agent evidence, channel-fit selection, qualified-demand accounting, distribution readiness, community-rule credibility, public data, developer tools, commerce, marketing, email, web operations, media, and productivity.
It audits caller-supplied release and discovery evidence for required surfaces, evidence freshness, name, version, canonical and price consistency, supported claims, and selected Show HN or Product Hunt launch constraints. It does not predict attention, demand, ranking, revenue, or platform acceptance.
It checks a proposed Reddit, Hacker News, Show HN, or Product Hunt action against maintained official guidance, caller-supplied Reddit community rules, account credibility evidence, promotion ratios, links, AI-text restrictions, and engagement-solicitation boundaries. It does not post or guarantee acceptance.
It separates exposure, operations, candidate signals, qualified demand, validated value, external revenue, and retention. Views, impressions, reactions, indexing, deployments, directory listings, scans, sent messages, and owner-funded verification do not count as demand or revenue.
It ranks supplied channel tests using audience fit, buyer intent, format, attribution, access, saturation, current rule evidence, product proof, budget, owner availability, a ready offer path, and a qualified outcome gate. It blocks vanity-only goals and does not publish or promise traction.
Version 0.5 added Chicago demolition-permit change packets, site-discovery change packets, CSV schema-drift preflights, release-artifact manifest validation, and email DNS authentication evidence checks.
Paid execution uses exact-price USDC payments through x402 on Base. Discovery, schemas, OpenAPI metadata, MCP tools, and sample fixtures are free.
No raw utility inputs or results are persisted. See the commercial telemetry document for the limited event data that may be recorded.
Yes. Read the live catalog and schemas, inspect the OpenAPI contract, enumerate MCP tools, and run against the published deterministic fixtures before authorizing a payment.
Start with the live catalog, choose one bounded product, and set the client’s maximum accepted price.