← CONNSKILL Growth Services
Conformance reports
for x402 sellers
A seller that accepts an invalid payment loses money. One that rejects a valid payment loses customers. This page documents HTTP refusals of altered, unsigned payment attempts. It does not establish isolated signature or replay security.
No valid payment is ever sent. Every request carries an authorization with exactly one field altered from the real challenge, plus a signature that is recognisably fabricated. It has to be refused. No money moves, on either side.
At most 11 requests per origin including the challenge, one per second, with a user agent that states its purpose and an opt-out address. Anyone who would rather not be checked writes to us and is removed.
Reports (228)
Evidence-bound summaries
A model selects and orders measured statements. Publication requires an exact match to the service, report and source fields. Held drafts remain private.
2026-10-22 · Source-linked summary
https://api.exa.ai: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://x402tap.com: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://api.gondola-ai.com: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://api.nansen.ai: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://lionx402.com: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://api.insumermodel.com: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://chat.gedx402.com: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://email-auth.use.x402atlas.com: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://api.polyform.org: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://horizonpulse.dev: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. These observations concern HTTP responses to unpaid or invalid payment requests. They do not prove isolated cryptographic or replay validation, successful delivery, or payment safety.
Inspect measurement evidence
2026-10-21 · Source-linked summary
https://glim.sh: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://api.aidress.ai: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://rpc.nodeflare.app: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://agent.pocket.network: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://x402.klimalabs.com: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://stabletravel.dev: 2 of 11 measured HTTP-response checks met the battery criteria; 9 did not. https://coil.trade: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://obol.danieldutoit.net: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://signals.nsgoods.org: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://anywaypossible.com: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. These observations concern HTTP responses to unpaid or invalid payment requests. They do not prove isolated cryptographic or replay validation, successful delivery, or payment safety.
Inspect measurement evidence
2026-09-15 · Source-linked summary
https://stableupload.dev: 2 of 11 measured HTTP-response checks met the battery criteria; 9 did not. https://api-dev-v2.ixs.finance: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://voidfeed.ai: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. These observations concern HTTP responses to unpaid or invalid payment requests. They do not prove isolated cryptographic or replay validation, successful delivery, or payment safety.
Inspect measurement evidence
2026-09-14 · Source-linked summary
https://laso.finance: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://cheaptokens.ai: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. https://google-trends.use.x402atlas.com: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. These observations concern HTTP responses to unpaid or invalid payment requests. They do not prove isolated cryptographic or replay validation, successful delivery, or payment safety.
Inspect measurement evidence
2026-09-11 · Source-linked summary
https://x402.ottoai.services: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. For https://api.hyperextend.xyz, the battery recorded 11 HTTP-response checks: 11 met its criteria and 0 did not. https://cars2026.vercel.app: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. These observations concern HTTP responses to unpaid or invalid payment requests. They do not prove isolated cryptographic or replay validation, successful delivery, or payment safety.
Inspect measurement evidence
2026-09-09 · Source-linked summary
https://api.bitrefill.com: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. For https://api.exa.ai, the battery recorded 11 HTTP-response checks: 11 met its criteria and 0 did not. https://api.nansen.ai: 11 of 11 measured HTTP-response checks met the battery criteria; 0 did not. These observations concern HTTP responses to unpaid or invalid payment requests. They do not prove isolated cryptographic or replay validation, successful delivery, or payment safety.
Inspect measurement evidence
Machine-readable summaries
What is tested
challenge_present — Does an unpaid request answer with a valid 402 challenge?
malformed_signature — Is an unreadable payment header refused rather than answered with a 5xx?
wrong_recipient — Is a payment to a different recipient refused?
wrong_asset — Is a token other than the one demanded refused?
underpayment — Is an amount below the price refused?
wrong_network — Is a payment on the wrong chain refused?
expired — Is an expired authorization refused?
replayed_nonce — Is a reused nonce refused?
not_yet_valid — Is a not-yet-valid authorization refused?
domain_version — Is a contract version supplied by the request ignored?
domain_chain_id — Is a chain id supplied by the request ignored?
Grading is strict: delivering despite an invalid payment is a finding, and so is a server error. Only a clean refusal in the 4xx range passes.
Fetch a report
curl -s https://agent.connskill.com/v1/conformance?origin=agent.connskill.com
Every report carries a fingerprint over its canonical form. Change the content and the fingerprint changes — and anyone who wants to recompute it can.