AgentBridge
AI knowledge marketplace for China data via x402 payments. Access university reports, industry briefings, and web content through Base chain USDC payments.
README
AgentBridge — Machine Commerce for China-Facing AI Agents
The internet was built for humans. APIs were built for software. But AI agents need something different: a way to discover, pay for, and consume digital capabilities autonomously.
AgentBridge is an experiment toward that future — building machine commerce infrastructure where AI agents can access specialized digital capabilities without accounts, subscriptions, or human intervention.
Not affiliated with other projects named "AgentBridge" ...
AgentBridge lets AI agents discover, purchase, and consume machine-readable digital capabilities related to China — through standardized APIs and x402 payments settled in USDC on Base. No API keys, no subscriptions, no account setup: an agent pays per call and gets a result.
Built on the x402 protocol, now stewarded by the Linux Foundation's x402 Foundation (members include AWS, Google, Visa, and Coinbase).
What It Provides
Capabilities are organized into two categories. Check which one you need before calling — they differ in speed, cost, and what kind of answer you get.
🔹 raw-content — unverified, machine-fetched, cheap
Fast, unreviewed raw content, priced in fractions of a cent. Use this when you just need to read what's currently published at a specific URL, as-is.
Reliability note: this returns the raw content as currently published at the source, at the moment of the fetch — unverified and not human-reviewed. If the source changes or is wrong, the result reflects that.
| Capability | Endpoint | What it does | Price |
|---|---|---|---|
| Web Fetch | POST /v1/fetch/dynamic |
Converts a Chinese webpage (Xiaohongshu, Zhihu, etc.) into clean Markdown/HTML. Handles JS rendering, anti-bot pages, and messy HTML that's normally hard for an agent to parse directly. | $0.003 (static) / $0.008 (dynamic) |
🔸 verified-analysis — human-authored, decision-grade
Slower to produce, priced higher, includes original comparative judgment — not just extracted facts. Use this when the task needs an answer you can act on, not just raw data.
| Capability | Endpoint | What it does | Price |
|---|---|---|---|
| University Employment Reports | POST /v1/reports/employment/{school_id} |
Human-authored deep-dive analysis of a Chinese university's officially published 2025 graduate employment report. | 90 USDC/report |
| University Teaching-Quality Reports | POST /v1/reports/teaching/{college_id}/{dimension} |
Same, for officially published teaching-quality reports, across 8 fixed analytical dimensions per institution. | 90 USDC/dimension |
| "15th Five-Year Plan" Strategic Notes | POST /v1/notes/15-5-package |
Human-authored analysis of China's 2026–2030 national plan as it relates to college application decisions. 41 articles. | 90 USDC/package |
Free full-length samples (not excerpts) for all verified-analysis products, plus the live, continuously updated catalog of available schools/colleges, are at the storefront.
More capabilities in both categories are added over time. The category a capability belongs to tells you what to expect from it — the list of specific products will keep growing.
How It Works
- Agent discovers the service via
/openapi.json,/llms.txt, or an x402 registry (see Status below). - Agent calls an endpoint without payment → gets
402 Payment Requiredwith an x402 payment challenge. - Agent pays in USDC on Base, retries with the payment header.
- Server verifies the payment against Coinbase's CDP facilitator, executes the request, and returns the result.
No API keys. No subscriptions. No account registration required for x402 payments.
When to Use Which
| If the task is... | Use | Why |
|---|---|---|
| "Get the current Zhihu Hot List / trending topics" | raw-content |
You just need what's on the page right now, no judgment needed on it. |
| "Summarize what's currently on this Xiaohongshu post" | raw-content |
Raw extraction of existing published content. |
| "Analyze the employment quality of a specific Chinese university" | verified-analysis |
Requires comparative judgment, benchmarking against peer institutions, and interpretation — not just numbers off a page. |
| "Should a student choose College A over College B based on teaching quality?" | verified-analysis |
A decision-grade answer, not a data dump. |
Quickstart for Agent Developers
# Calling the raw-content fetch endpoint (returns 402 if unpaid)
curl -i -X POST https://api.060504.shop/v1/fetch/dynamic \
-H "Content-Type: application/json" \
-d '{"url": "https://www.zhihu.com/question/xxx"}'
An unpaid call returns 402 Payment Required with the payment challenge in the PAYMENT-REQUIRED response header (base64-encoded x402 v2 object). Decode it, pay the required amount in USDC on Base, and retry the request with your X-Payment header set to the signed payment payload — see the x402 specification for the full handshake, or the full OpenAPI spec for this service's exact schemas.
import requests
url = "https://api.060504.shop/v1/fetch/dynamic"
payload = {"url": "https://www.zhihu.com/question/xxx"}
res = requests.post(url, json=payload)
if res.status_code == 402:
payment_required = res.headers.get("PAYMENT-REQUIRED") # base64-encoded x402 v2 challenge
# decode it, sign and settle payment with your Base/USDC wallet per the x402 spec,
# then retry the POST with an X-Payment header carrying the signed payload
Agent Integration
AgentBridge Atlas is designed for the emerging agent economy.
AI agents can discover and consume Atlas capabilities through machine-readable interfaces.
Supported integration layers:
- OpenAPI — standard API access
- x402 — machine-native payments with USDC on Base
- MCP — agent tool integration
- Agent discovery metadata — machine-readable identity
Resources:
-
Agent discovery:
.well-known/agent.json -
MCP adapter:
/mcp -
Claude Desktop example:
/examples/claude_desktop_config.json
The goal is to make AgentBridge Atlas discoverable and usable by autonomous AI agents and agent frameworks.
Status
- ✅ Live API:
https://api.060504.shop - ✅ Listed on 402 Index
- ✅ Discoverable via
/.well-known/agent.json(ReqCast-compatible) - ✅ On-chain payment history is fully public and verifiable: view on BaseScan
Quick Links
- API Documentation (openapi.json)
- Storefront & Free Samples
- 402 Index Listing
- Legal Disclaimer — by using this service, you agree to its terms
- AgentBridge Atlas Vision
Writing / Background
- AgentBridge: Building a Pay-Per-Fetch Gateway for AI Agents on Base
- How AI Agents Can Legally Access Chinese Public Data
- AgentBridge: The Missing Data Layer for AI Agents Targeting China
- How We Built a Pay-Per-Report Data Service for 72 Chinese Universities Using x402
- More on Medium
Two categories, one platform: raw web content and human-authored analysis, both discoverable and payable the same way.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.