5 x402 Endpoints Your Claude Code Agent Will Actually Use

August 10, 2026 · 4 days until Claude Code auto mode ships as default

On August 5, Solana and Coinbase ran the "Agentic Payments" webinar. The demo was simple: an agent was asked for the day's top tokens by trading volume. It found the right x402 endpoint, paid a one-cent fee, and returned live rankings — no API key, no signup, no human in the loop. Then it chained another call: trending social content → video generation endpoint → market recap clip. The agent funded and paid for the entire workflow autonomously.

In 4 days, Claude Code auto mode becomes the default for Pro, Max, and Team users. Every developer's agent will be able to encounter an HTTP 402, read the payment terms, and decide whether to pay — the same way the Solana demo agent did. Here are the 5 endpoints your agent is most likely to chain together, with working curl commands you can test right now.

1. 🔍 x402-recall most-used

1,803 trials · 98 agents · ~$0.005/call

Web search with AI summarization. When your agent needs to answer "what happened in the last 24 hours?" — this is the first call in the chain. It's the top endpoint by trial volume because every research task starts with "find the information."

curl -s "https://minia2a.uk/x402/recall?q=latest+Claude+Code+updates&trial=1"

Agent flow: "Research X" → x402-recall finds sources → x402-fetch reads them → x402-summarize condenses → response to user. The first link in every research chain.

2. ⛽ x402-gas 2nd most-used

1,305 trials · 174 agents · ~$0.003/call

Real-time gas prices across Ethereum, Base, Arbitrum, Optimism, and Polygon. 174 agents use this — the broadest adoption of any endpoint. Your agent needs gas prices before submitting any transaction, making this the most common pre-flight check.

curl -s "https://minia2a.uk/x402/gas?trial=1"

Agent flow: Before any on-chain action → check gas → if <threshold, proceed → submit tx. Agents use this like humans use gas trackers — compulsively, before every move.

3. 🤖 x402-captcha-solve high-volume

1,202 trials · 133 agents · ~$0.01/call

Solve image and audio CAPTCHAs programmatically. The sleeper hit of the marketplace — 133 agents use it because web scraping hits CAPTCHA walls constantly. Every autonomous agent that browses the web will eventually need this.

curl -s "https://minia2a.uk/x402/captcha-solve?trial=1"

Agent flow: Scrape site → hit CAPTCHA → x402-captcha-solve → resume scraping. The invisible infrastructure of autonomous browsing.

4. 🔎 x402-find discovery

1,167 trials · 84 agents · ~$0.003/call

Discover x402 endpoints by capability. This is the meta-endpoint — your agent uses it to find other endpoints. When an agent encounters a task it doesn't have a tool for, it calls x402-find to discover one that can handle it.

curl -s "https://minia2a.uk/x402/find?q=token+price+ethereum&trial=1"

Agent flow: "I need a price oracle" → x402-find("token price") → discover x402-price-oracle → call it. The discovery layer in action.

5. 📸 x402-screenshot visual

332 trials · 24 agents · ~$0.02/call

Capture webpage screenshots. Lower volume than the top 4, but highest stickiness — agents that use it keep using it. Essential for visual verification tasks: "does the deployed site look right?", "show me the dashboard", "capture the error state."

curl -s "https://minia2a.uk/x402/screenshot?url=https://example.com&trial=1"

Agent flow: Deploy → x402-screenshot to verify → compare with expected → report. Visual feedback loop for autonomous ops.

🔗 The Full Chain: What Auto Mode Actually Looks Like

This is exactly what the Solana/Coinbase webinar demoed — and exactly what your Claude Code agent will do on Aug 14:

x402-find("crypto market data") → discovers available endpoints → $0.003
x402-recall("today top performing tokens") → gets the raw data → $0.005
x402-fetch(urls from recall) → pulls detailed pages → $0.005
x402-screenshot(dashboard URL) → captures visual state → $0.02
Total: ~$0.033 for a fully autonomous research + verification workflow

The agent decides at each step: is the next call worth it? With a .agent-budget file set to $1/day, this entire chain costs 3.3% of the daily budget. The economics already work.

What This Means for API Builders

The Solana webinar proved the technical flow works. Claude Code auto mode will put it in millions of developer hands. If you're building an API that agents might need, the checklist is short:

Return HTTP 402 with machine-readable headers: x-402-amount, x-402-chain, x-402-token, x-402-recipient. Without these, auto-mode agents can't decide whether to pay.
Offer free trials: Agents discover through experimentation. If your first call requires payment, you're invisible to the discovery layer. Trial-first is how 181 endpoints with fewer than 20 trials still get discovered.
Be discoverable: Register in at least one discovery index. An endpoint that no agent can find doesn't exist, no matter how good the API is.

The Bigger Picture: 95% Signaling, 5% Real Commerce

The Forkast article published this week notes that over 95% of x402's 200 million transactions are protocol signaling — machines testing the plumbing, not buyer-seller exchange. Real daily commercial volume was approximately $28,000 as of March. But this isn't a failure mode; it's what a pre-adoption phase looks like. The infrastructure tests itself until the applications arrive.

Claude Code auto mode might be the first application that converts signaling into commerce at scale. Millions of developers, each with an agent that can autonomously encounter and pay for API calls. The plumbing is tested. The endpoints are live. The remaining question is discovery — which of the 150,000 x402 endpoints does your agent actually need?

← Back to minia2a · Browse all endpoints →