The Problem
Every AI agent that calls external services — inference, translation,
data, images, audio — requires the developer to manually find each API,
sign up, manage keys, write integration code, and handle billing
separately for every provider.
One agent workflow can touch a dozen services. That doesn't scale.
The Solution
Wayforth. One install:
uvx wayforth-mcp
Then two tool calls from any agent:
# Discover
wayforth_search("translate text to Spanish")
# → DeepL API WRI: 82 Tier 2 Verified $0.0000025/req
# → LibreTranslate WRI: 71 Tier 2 Verified Free
# → ModernMT WRI: 68 Tier 2 Verified $0.000003/req
# Pay
wayforth_pay(service_id, owner_address, amount_usdc=0.001)
# → Non-custodial calldata. Settles on Base in ~2 seconds.
No API keys. No billing relationships. No integration code.
Works in Claude Code, Cursor, Windsurf, and any MCP-compatible runtime.
What's Live
200+ real API endpoints across 7 categories — inference, data,
translation, image, code, audio, embeddings.
165+ Tier 2 verified services — automatically probed every 6 hours,
90%+ uptime required, auto-demoted after 3 consecutive failures.
No paid placement. Ever.
WayforthRank — a proprietary ranking engine combining semantic
relevance, reliability history, and real agent payment conversion
signals. Rankings improve with every query. (Patent pending)
Smart contracts on Base Sepolia — non-custodial escrow, audited,
Basescan verified. Mainnet Q3 2026.
Tiered routing fee 0.75%–1.5% — the only cost. No fixed fee.
Viable for sub-cent micro-transactions.
The Architecture
Four layers:
- Settlement Layer — Base blockchain, WayforthEscrow, USDC
- Verification Layer — Coverage tier system, automated probing
- Intelligence Layer — WayforthRank, Service Graph, Wayforth Identity
- Agent Layer — MCP server, Python SDK, TypeScript SDK, WayforthQL
WayforthRank — How Ranking Works
WRI (Wayforth Reliability Index) is a score from 0–100 combining:
| Signal | Points |
|---|---|
| Base score | 50 |
| Tier 2 verified | +20 |
| Zero failures | +10 |
| Probed within 24h | +10 |
| x402 protocol | +5 |
| Agent popularity (7d) | +5 max |
| Payment conversions (7d) | +8 max |
The payment conversion signal is the strongest — it only accumulates
from real agent payments. A competitor needs 12+ months of real agent
payment data to replicate it.
The Data Flywheel
Every search query links to a query_id. When an agent pays for a
service, that payment converts the search record and feeds WayforthRank.
Rankings improve with every real transaction.
WayforthQL — Structured Queries
For agents that need more control:
POST /query
{
"query": "fast inference for coding agents",
"tier_min": 2,
"protocol": "x402",
"sort_by": "wri",
"price_max": 0.001,
"limit": 5
}
Developer Dashboard
Every developer gets a personal dashboard at wayforth.io/dashboard:
- API key management
- Usage and quota tracking
- Search history and analytics
- Service reliability trends
- Agent identity trust scores
Free tier — 1,000 searches/month. No credit card required.
Try It
uvx wayforth-mcp
- Website: https://wayforth.io
- GitHub: https://github.com/WayforthOfficial/wayforth (BSL 1.1)
- PyPI: https://pypi.org/project/wayforth-mcp/
- Quickstart: https://wayforth.io/quickstart
- Whitepaper: https://wayforth.io/wayforth-whitepaper-v3.pdf
Happy to answer questions about the WayforthRank architecture,
payment routing design, or the coverage tier system.





Top comments (1)
Update — catalog expanded significantly since publishing:
We've grown from 190+ to 270+ real API endpoints and added 10 new
categories since this post went live.
New categories now in the catalog:
Full catalog now: 270+ real APIs · 230+ Tier 2 verified · 18 categories
wayforth.io/leaderboard to see what agents are searching most.