openapi: 3.2.0 info: title: LEGIT Stats API description: Trust-check any x402 merchant or agent address on Algorand, Base, Solana, or 17 other chains before routing a payment. version: 0.1.0 contact: name: GONNA url: https://gonna.bond/gonnaverse/ email: legit@gonna.bond x-logo: url: https://legit.gonna.bond/favicon.png x-guidance: 'LEGIT is the trust and routing layer for the x402 agentic economy. Trust-check any x402 merchant address on Algorand, Base, Solana or 17 other chains before routing a payment: grades (A+ to F) and scores (0-100) come from live uptime, latency and payment-coherence probes, never self-reported data. Run as a pre-payment trust check: call GET /v1/check/{address} before any x402 payment to an address you have not paid before. Start FREE: GET /v1/check/{address} (full trust report), /v1/leaderboard (ranked shortlist), /v1/trending, /v1/pulse, /v1/arena/preview. Paid per call in USDC over x402: GET /v1/compare (2-5 merchants head-to-head), /v1/arena (24h cross-chain settlement benchmark, ?network= filter), /v1/history?address=... (30-day trust trend), POST /v1/watch (7- or 30-day change sentinel, webhook-armed), POST /v1/deep-check (due diligence, up to 10 addresses). How to pay: call a paid endpoint without paying and it answers 402 with a PaymentRequired document listing every active rail; pay one rail and retry with the PAYMENT-SIGNATURE header. Rails cheapest first: algorand-mainnet $0.004, base-mainnet and solana-mainnet $0.005 per call. Watch companions GET /v1/watches/{watch_id} and /v1/watches/{watch_id}/events are FREE: the watch id is the bearer capability for status and the event feed. Webhook payload anatomy: /docs; agent manifest: /llms.txt. Use LEGIT before sending an x402 payment to an unknown merchant, before accepting a counterparty in agent-to-agent settlement, or to monitor a critical upstream for trust degradation.' tags: - name: Stats paths: /v1/stats: get: summary: Read headline service stats and storage facts description: 'Headline service stats: merchant and measurement counts, average trust score and verifiable storage facts (FREE, cached 30s).' operationId: stats_v1_stats_get responses: '200': description: Successful Response content: application/json: schema: $ref: '#/components/schemas/Stats' security: [] tags: - Stats components: schemas: Stats: properties: merchants: type: integer title: Merchants merchants_testnet: type: integer title: Merchants Testnet default: 0 merchants_mainnet: type: integer title: Merchants Mainnet default: 0 merchants_active: type: integer title: Merchants Active default: 0 measurements: type: integer title: Measurements avg_score: type: number title: Avg Score storage: $ref: '#/components/schemas/StorageInfo' type: object required: - merchants - measurements - avg_score - storage title: Stats StorageInfo: properties: path: type: string title: Path persistent: type: boolean title: Persistent db_bytes: anyOf: - type: integer - type: 'null' title: Db Bytes retention_days: anyOf: - type: integer - type: 'null' title: Retention Days tables: anyOf: - type: object - type: 'null' title: Tables tables_bytes: anyOf: - type: string - type: 'null' title: Tables Bytes type: object required: - path - persistent title: StorageInfo description: 'Wave 8.6: verifiable storage facts (resolved SQLite path + whether it lives on the persistent volume). Wave 40 (The Diet): disk observability -- ``db_bytes`` is the on-disk size of the DB file (null when unreadable) and ``retention_days`` the raw-measurement retention window actually in force (older history survives as daily rollups, not raw rows). Wave 41 (La Bellezza, W6): growth introspection -- ``tables`` maps each main table to its real row count (plus ``bytes`` when SQLite''s dbstat virtual table is compiled in). ``tables_bytes`` names the bytes source ("dbstat") or is null, in which case entries honestly carry row counts only. Wave 43 (La Grande Dieta, D5): table entries additionally carry ``index_bytes`` (that table''s index pages via dbstat, additive) when dbstat is available -- ``bytes`` alone under-reported the real on-disk cost per table by 1.4x-2.9x.'