DEX Data & Agent Utilities

active

Pay-per-call (x402/USDC on Base) endpoints: SEC EDGAR filings, US Treasury yield curve, Stablecoin supply, Entity and compliance lookups, Agent utilities, x402 seller checks and agent demand index.

AIBasex402 v2exactdonnyautomation.com ↗︎

Settled via Coinbase.

Transactions · 30d
4
Volume · 30d
$0.12
Unique buyers · 30d
1
Uptime · 30d
30.2%
Latency p50
99ms
Reported calls · 30d
9

Endpoints (24 live)

  • GET /treasury — US Treasury par yield curve, every constant maturity from 1 month to 30 years for a published business day. Returns curve[] with tenor, months and percent, spreads (2s10s, 3m10y, 5s30s) and an explicit inversion flag - the real question is whether the curve is inverted, not what fourteen numbers are. A tenor Treasury did not publish is null, never zero. Pass ?latest=1 OR ?date=YYYY-MM-DD, not both. Errors: 400 conflicting_params|bad_latest|bad_date, 404 no_curve with the available range. (0.03 USDC on Base)
  • GET /edgar/events — Form 8-K material events for one issuer SINCE a cursor, each item code decoded to the title SEC prints (2.02 Results of Operations, 5.02 Departure of Directors, 1.05 Cybersecurity Incidents). Requires ?since=YYYY-MM-DD AND ?ticker= or ?cik=. Optional ?items=1.01,5.02 (matches ANY), ?limit=1..200. Errors: 400 missing_since|missing_issuer|bad_items, 404 unknown_ticker|unknown_cik, 502 edgar_unavailable, 503 rate_limited_locally. AN EMPTY DELTA IS CHARGED. Every form: /edgar/filings. (0.1 USDC on Base)
  • GET /edgar/filings — Every SEC filing by one issuer SINCE a cursor - a delta, not a dump. Requires ?since=YYYY-MM-DD AND ?ticker= or ?cik= (cik wins if both). Optional ?forms=10-K,8-K (also matches /A), ?limit=1..200. Errors: 400 missing_since|missing_issuer|bad_since|bad_ticker|bad_cik|bad_limit, 404 unknown_ticker|unknown_cik, 502 edgar_unavailable, 503 rate_limited_locally. AN EMPTY DELTA IS CHARGED: nothing-changed is the answer you bought. 8-K with decoded items: /edgar/events. Ownership: /edgar/insiders. (0.1 USDC on Base)
  • GET /feed — RSS, Atom or RDF feed to clean JSON. Returns feed (title, link, description, updated) and items[] with title, real destination link, id, author, categories and the date both verbatim and ISO 8601, plus each summary twice: feed HTML and plain text. An unparseable date is null, never invented. Requires ?url=; optional ?limit=1..100 (default 20). Errors: 400 missing_url, 403 blocked_private, 422 not_xml|malformed_xml|not_a_feed. For one article body use /markdown. (0.01 USDC on Base)
  • GET /markdown — Fetch a public article or PDF and return clean Markdown plus title, byline, siteName, excerpt and wordCount. HTML is extracted with Firefox reader-mode rules; PDFs return their text layer. Requires ?url=<public http(s) URL>. Errors: 400 missing_url|bad_url, 403 blocked_private (private and internal hosts refused, every redirect hop re-checked), 413 too_large above 8 MB, 422 no_text_layer for scanned PDFs or not_extractable for app shells, 504 fetch_timeout. (0.01 USDC on Base)
  • GET /edgar/insiders — Ownership changes for one issuer SINCE a cursor: Forms 3/4/5 (Section 16 officers, directors) and Schedules 13D/13G (5%+ holders), amendments included. Returns the FILING-level change and its document URL; share counts inside each Form 4 are NOT parsed and not claimed. Requires ?since=YYYY-MM-DD AND ?ticker= or ?cik=. Optional ?forms=4, ?limit=1..200. Errors: 400 missing_since|missing_issuer, 404 unknown_ticker, 502 edgar_unavailable. AN EMPTY DELTA IS CHARGED. 8-K: /edgar/events. (0.1 USDC on Base)
  • GET /lei — Legal Entity Identifier lookup from the GLEIF golden copy, searchable by NAME - knowing the LEI already is the hard part. Returns records[] with lei, legalName, status, jurisdiction, legalAddress and a lapsed flag; LAPSED and RETIRED entities are returned, not hidden - a filtered record and no record look identical. Requires ?q=<legal name, max 200> or ?lei=<20 chars>. Errors: 400 missing_query|bad_lei, 404 no_match, 502 lei_unavailable. Sanctions screening is /sanctions. (0.03 USDC on Base)
  • GET /random — Cryptographically secure randomness for agents that are deterministic or sandboxed. Two modes: raw bytes (?bytes=1..1024, default 32, clamped) returning hex and base64url; or uniform integers (?min= and ?max= together, optional ?count=1..1000) rejection-sampled so there is no modulo bias. Errors: 400 bad_bytes for a non-integer, bad_range when max is not greater than min, range_too_large. Generated server-side, and the response says so - if nobody may have seen it, generate it yourself. (0.005 USDC on Base)
  • POST /ai/extract — Pull named fields out of your text with an AI model (Claude Haiku 5.5). POST JSON {text, fields}: text max 32,000 bytes; fields 1..10 names of at most 40 chars, else 400 bad-fields. Returns {values: {field: string or null}}; null when the text does not say. Errors: 400 bad-json|unknown-field|missing-text, 413 text-too-long, 422 refused, 429 rate-limited, 502 model-upstream, 503 model-budget-exhausted. ERRORS ARE NOT CHARGED. Whole-text summary: /ai/summarize. Pick one label: /ai/classify. (0.0085 USDC on Base)
  • POST /ai/rephrase — Rewrite your own text in one style with an AI model (Claude Haiku 5.5). POST JSON {text, style}: text max 32,000 bytes; style plain|formal|shorter|bullets, else 400 bad-style. Returns {text}, facts kept. Errors: 400 bad-json|unknown-field|missing-text, 413 text-too-long, 422 refused, 429 rate-limited, 502 model-upstream, 503 model-budget-exhausted. ERRORS ARE NOT CHARGED. Shorter by sentence count: /ai/summarize. Fields: /ai/extract. (0.0085 USDC on Base)
  • POST /ai/classify — Put your own text in one of YOUR labels with an AI model (Claude Haiku 5.5). POST JSON {text, labels}: text max 32,000 bytes; labels 2..10 names of at most 40 chars, else 400 bad-labels. Returns {label, confidence: high|medium|low}. Errors: 400 bad-json|unknown-field|missing-text, 413 text-too-long, 422 refused, 429 rate-limited, 502 model-upstream, 503 model-budget-exhausted. ERRORS ARE NOT CHARGED. Free-form fields: /ai/extract. Summary: /ai/summarize. (0.0085 USDC on Base)
  • POST /ai/summarize — Summarize your text with an AI model (Claude Haiku 5.5). POST JSON {text, sentences?}: text required, max 32,000 bytes; sentences 1..10, default 3, else 400 bad-sentences. Returns {summary, model, input_tokens, output_tokens, maxTokensHit, notice}. Errors: 400 bad-json|unknown-field|missing-text, 413 text-too-long, 422 refused, 429 rate-limited, 502 model-upstream, 503 model-budget-exhausted. ERRORS ARE NOT CHARGED. Named fields: /ai/extract. One label: /ai/classify. Restyle: /ai/rephrase. (0.0085 USDC on Base)
  • GET /ai/extract (0.0085 USDC on Base)
  • GET /ai/rephrase (0.0085 USDC on Base)
  • GET /ai/classify (0.0085 USDC on Base)
  • GET /ai/summarize (0.0085 USDC on Base)
  • GET /sanctions — OFAC screening against the Treasury SDN and Consolidated lists - 19,600+ parties, 40,000+ names including aliases. Returns match, riskLevel, matches[] with score, matchedName, matchedOn and programs, plus screened and notScreened naming what was NOT checked (EU, UK OFSI, UN). Word order does not matter: OFAC stores people SURNAME, Given. Requires ?name= (max 200); optional ?minScore, ?type, ?limit. Errors: 400 missing_name|unsearchable_name, 503 sanctions_list_unavailable - never a false clear. (0.01 USDC on Base)
  • GET /federal-register (0.005 USDC on Base)
  • GET /edgar/dossier (0.03 USDC on Base)
  • GET /stablecoin/supply (0.008 USDC on Base)

+4 more endpoints.

First seen · last seen · last active