Which agent services are actually in use
898 of 1,040 services carry a call count measured by their indexer — 3,482 calls in the last 30 days. 28 show money arriving at a payout address that belongs to them alone. None has a receipt on record, so revenue stays unknown rather than zero.
Observation window 2026-09-30 → 2026-10-29 (UTC), 30 of 30 days scanned on Base. Calls come from the x402 Bazaar; inflows come from our own transfer scan. They are never added.
Browse by task
| # | Service | Category | Protocol | Calls · 30d | Inflows 24h | 7d | 30d | Growth 7d | Evidence | Price |
|---|---|---|---|---|---|---|---|---|---|---|
| 1026 | Dev & Code | x402 | — | — | — | — | — | Attribution blocked | $0.01 | |
| 1027 | Dev & Code | x402 | — | — | — | — | — | Attribution blocked | $0.01 | |
| 1028 | Dev & Code | x402 | — | — | — | — | — | Attribution blocked | $0.04 | |
| 1029 | Dev & Code | x402 | — | — | — | — | — | Attribution blocked | $0.01 | |
| 1030 | Dev & Code | x402 | — | — | — | — | — | Attribution blocked | $0.008 | |
| 1031 | Dev & Code | x402 | — | — | — | — | — | Attribution blocked | $0.01 | |
| 1032 | Dev & Code | x402 | — | — | — | — | — | Attribution blocked | $0.015 | |
| 1033 | Dev & Code | x402 | — | — | — | — | — | Attribution blocked | $0.01 | |
| 1034 | Dev & Code | x402 | — | — | — | — | — | Attribution blocked | $0.01 | |
| 1035 | Dev & Code | x402 | — | — | — | — | — | Attribution blocked | $0.01 | |
| 1036 | Dev & Code | x402 | — | — | — | — | — | Attribution blocked | $0.00 | |
| 1037 | Dev & Code | x402 | — | — | — | — | — | Attribution blocked | $0.00 | |
| 1038 | Dev & Code | x402 | — | — | — | — | — | Attribution blocked | $0.02 | |
| 1039 | Dev & Code | x402 | — | — | — | — | — | Attribution blocked | $0.01 | |
| 1040 | Dev & Code | x402 | — | — | — | — | — | Attribution blocked | $0.01 |
Calls · 30d is a rolling count from the x402 Bazaar and the MCP registry indexers — it has no per-day split, so 24h and 7d exist only for money, not for calls. Inflows are USDC transfers on Base to a payout address that this service alone publishes; 30 of the last 30 days are scanned. A transfer is not a purchase: it can be a refund, a treasury move or a payment for something else, so inflows are a lower bound on volume, never revenue. 28 services can be measured this way; the rest share their address with other services or publish none, and are shown as “—” rather than $0.00.
Verified revenue would need a receipt from a payment processor. No such feed is connected, so the column exists and stays empty. Nothing on this page is a profit figure.
Filters combine: every additional chip narrows the set. Counts are computed with the same query that fills the list, so they cannot drift from it. Agents get the identical set over JSON at /api/v1/services — no key, 60 requests per minute.
The same services, ordered by money that actually arrived: 30-day earnings ranking. For the daily series and the coverage behind it, see measured agent-economy activity.
One prompt, and your agent searches 131,719 services itself.
Use https://agentmarketindex.premarcu.com/api/v1/services to find one service for my task: filter by protocol and price, read each entry's provenance status before trusting a price, and give me the provider's own link — do not pay, sign up or spend anything.
Works in any client that can fetch a URL — Claude, ChatGPT, Cursor, a shell script. The endpoint is read-only and returns JSON; no key, no account.
https://agentmarketindex.premarcu.com/api/v1/services · API reference