Bezahlte Endpunkte für autonome Agenten. Abrechnung über x402 in USDC auf Base Mainnet.
POST /zap-radar
Collects kind 9735 zap receipts addressed to one public key over up to twelve relays, merges them by event id and reports who sent how much over the requested window. Every amount comes from decoding the BOLT-11 invoice carried in the receipt, not from the amount tag of the zap request, because the two can disagree and only the invoice was payable. Each receipt is checked in three ways and the counts are reported: the receipt signature per BIP-340, whether the invoice description hash equals SHA-256 of the description tag holding the zap request, and whether the zap request itself carries a valid signature. Receipts failing any of these are counted separately and excluded from the totals, so a sum is never inflated by unverifiable claims. The answer also names every relay queried, whether it answered, and how many receipts each contributed — a total from four of ten relays is not the same statement as a total from ten. It does not prove payment settled on the Lightning Network; a receipt is issued by the recipient LNURL service and is evidence, not proof.
0.004 USDC je Aufruf · MCP nostr_zap_radar
POST /event-sign
Three modes over the same event structure. In preimage mode the caller supplies a public key and the event fields; the answer holds the exact NIP-01 serialisation, the SHA-256 event id derived from it and the 32 bytes to sign, so the caller signs locally and no secret key ever crosses the connection. In ephemeral mode a fresh secp256k1 keypair is generated for this single call, the event is signed with it, and the complete event plus its secret key are returned once and never stored — the intended use is a throwaway publishing identity such as an RSS bridge or an announcement bot, not a durable account. In verify mode a complete event is checked and each result is reported separately: whether the fields are well formed, whether the id matches the serialised content, and whether the Schnorr signature matches id and public key, because a wrong id and a wrong signature have entirely different causes. This service never accepts a foreign secret key in any mode; an nsec passed anywhere is refused with an explicit error rather than silently processed.
0.001 USDC je Aufruf · MCP nostr_event_sign
POST /relay-health
Every relay is measured twice over. Over WebSocket a genuine REQ query is sent and the connection time, the time to end of stored events and the number of events returned are recorded — a relay that accepts the connection and then stays silent is the failure mode a reachability ping cannot see. Over https the NIP-11 information document is read on the same host, yielding software, version, supported NIPs and the declared limits including whether the relay requires payment, requires authentication or restricts writes. Both results are combined into one verdict per relay: healthy, healthy_paid, reachable_but_empty, answers_without_eose, auth_required or unreachable, and the reason is always stated in the same record. Relays are measured in parallel, so the answer time is that of the slowest relay, not their sum. This is a point measurement at one moment from one location: it is evidence about now, not an uptime history, and a relay unreachable from here may be reachable elsewhere.
0.003 USDC je Aufruf · MCP nostr_relay_health
POST /rss-to-nostr
Fetches the feed over https with redirect and size limits, recognises RSS 2.0, RDF and Atom, and converts each entry into a NIP-01 event. Kind 1 produces a short note built from title, optional shortened summary and the link; kind 30023 produces a long-form article carrying the full entry text plus the title, published_at and d tags that NIP-23 expects. Every event gets an r tag with the entry URL and a t tag per feed category, so the result is addressable on the network rather than being plain text. Given a public key the exact event id is computed for each event and returned with it, which leaves the caller nothing to do but sign 32 bytes locally; without a key the events come back without ids because an id depends on the signing key. The events are never signed here and no secret key is accepted. created_at follows the entry publication date when the feed provides a parseable one, and the answer states per entry whether it did, because a feed without dates otherwise silently produces events all stamped with the fetch time.
0.002 USDC je Aufruf · MCP nostr_rss_to_events
POST /nip05-verify
Fetches https://domain/.well-known/nostr.json with the name query parameter and reads the mapping the domain publishes. The answer states the resolved public key in hex and as npub, the relay hints the document lists for that key, and the HTTP status and final URL after redirects, so a redirect to a login page is distinguishable from a genuine answer. Given an expected public key, the comparison is made and the result is reported as an explicit match or mismatch rather than left to the caller. A bare domain is treated as the root identifier _@domain, exactly as NIP-05 prescribes. The local part is compared case-insensitively against the document keys and the answer says which spelling the document actually used, since that difference is a frequent cause of a verification that fails for no visible reason. What this proves is narrow and stated in the answer: that the domain currently maps this name to this key. It does not prove who controls the key, and it is worth exactly as much as control over the domain.
0.001 USDC je Aufruf · MCP nostr_nip05_verify
POST /follower-graph
Fetches kind 3 contact lists over up to twelve relays. Because kind 3 is replaceable, several relays regularly hold different versions; only the one with the highest created_at is used and the answer states which relays carried that version and how many older versions were seen, so a stale relay never quietly overwrites a current list. At depth 1 the answer is the account and its outgoing follows. At depth 2 the contact lists of up to twenty-five contacts are fetched as well, which yields the edges between them, an overlap count per contact, and the mutual follows — the pairs that follow each other, which is the only reciprocity a contact list can actually establish. The result is a plain node and edge structure suitable for direct rendering, with degree counts already computed. One limitation is stated in every answer and cannot be worked around: a contact list says who someone follows, never who follows them. Inbound follows can only be approximated within the fetched neighbourhood, and the answer reports it as such rather than as a follower count.
0.004 USDC je Aufruf · MCP nostr_follower_graph
POST /lnurl-invoice
Accepts three forms of recipient. A bech32 lnurl is decoded to its callback URL, a Lightning address is resolved through the well-known lnurlp path, and a Nostr public key is resolved by reading its kind 0 profile from relays and taking the lud16 or lud06 field it publishes. The pay request is fetched, its declared limits are checked against the requested amount before any callback is made, and the answer reports minSendable, maxSendable, the allowed comment length and whether the service accepts Nostr zaps. The invoice returned by the callback is then decoded and checked rather than passed through: the invoice amount must equal the requested amount, the description hash must equal SHA-256 of the metadata the service published, and the expiry is resolved into an absolute timestamp, since an invoice without an explicit expiry field is valid for one hour and not indefinitely. Any mismatch is reported as a named finding and the invoice is still returned, so the caller decides rather than the service. A pre-signed NIP-57 zap request can be passed through for a zap; it is checked for a valid signature first and never created here, because creating one would require a secret key. No payment is made by this endpoint: it produces an invoice, nothing more.
0.002 USDC je Aufruf · MCP nostr_lnurl_invoice
POST /key-convert
Takes any public NIP-19 identifier or a 64-character hex value and returns every representation it can be expressed as. A decoded nprofile or nevent yields not only the key or event id but also the relay hints, the author and the kind that the TLV form carries — precisely the fields that make a follow-up query targeted, and precisely what is lost by code that only handles npub. Bare hex is ambiguous by nature, since the same 32 bytes are a valid public key and a valid event id; both readings are returned and labelled rather than one being guessed. Relay hints, author and kind can be supplied to build a richer identifier, and every relay address given is validated as a wss address before it is embedded. Secret keys are refused in every form: an nsec passed here yields an explicit error rather than a conversion, because a secret key on a network connection cannot be un-sent. This is a pure conversion with no network access, so it is exact rather than a best effort.
0.001 USDC je Aufruf · MCP nostr_key_convert
POST /bot-score
Fetches the recent notes of one public key across several relays and measures six independent signals: posting rate per day, the coefficient of variation of the intervals between posts, how many of those intervals sit within a few seconds of each other, the entropy of the hour-of-day distribution, the share of posts whose text repeats an earlier post verbatim, and the largest burst inside sixty seconds. Each signal has a fixed threshold, and the answer names for each one its measured value, its threshold and the points it contributed, so the total can be recomputed and disputed rather than merely believed. Below eight events no score is produced at all — the timing signals have no meaning on a handful of posts, and a confident number there would be an invention. The result is a behavioural indication, not a verdict on an account and not evidence of spam: legitimate announcement bots score high by design and a careful spammer scores low. Only the shape of the activity is examined, never the content, and the relay sample is finite, so the answer reports how many relays answered and whether the fetch limit was reached.
0.004 USDC je Aufruf · MCP nostr_bot_score
POST /zap-receipt-check
Takes a kind 9735 receipt directly or fetches it from relays by event id, then runs eight named checks and reports each one separately with its reason. The receipt signature and the signature of the zap request enclosed in the description tag are verified per BIP-340. The invoice is decoded and its amount is compared against the amount tag of the zap request, since those two can disagree and only the invoice was payable. The invoice description hash is compared against SHA-256 of the description tag, which is the link that ties a specific invoice to a specific zap request; a receipt failing this check carries an invoice that was never issued for it. Recipient and sender in the p and P tags are compared against the zap request. Optionally the recipient LNURL service is read and the receipt issuer is compared against the nostrPubkey it publishes — the check that actually distinguishes a genuine receipt from one signed by any key at all, and the one most implementations omit. The result is a verdict with the failed checks named. One thing is stated plainly in every answer: a zap is a Lightning payment and never touches the Bitcoin chain, so no on-chain confirmation exists or can be produced, and a receipt is evidence issued by the recipient service rather than proof of settlement.
0.002 USDC je Aufruf · MCP nostr_zap_receipt_check