Agent Intent Protocol
The open infrastructure standard for the agentic web.
A dual-sided protocol for machine-readable routing (jin.json) and zero-latency perimeter security.
{
"aip_version": "0.2.2",
"app": {
"name": "Sovereign API Gateway",
"description": "Self-governing endpoints with cryptographic validation"
},
"security": {
"shield": "enabled",
"provider": "asymmetric-rs256"
},
"intents": [
{
"id": "fetch_secure_data",
"method": "POST",
"endpoint": "/api/v1/secure-fetch"
}
]
}The Sovereign Agentic Economy
Choose your side of the protocol.
Jin aligns the incentives of webmasters and developers by providing a unified trust and capability standard.
Webmaster Sovereignty
Take back your traffic. For years, websites have been forced to fight hostile, brute-force AI scrapers draining their servers. Jin gives you the power back.
- Stop Blind Scraping: Block aggressive scrapers instantly at the perimeter gateway before they hit your compute.
- Reduce Server Load: Short-circuit unauthorized requests at the door with light, high-performance in-memory middleware validation.
- Control Perimeter Boundaries: Establish the exact perimeter boundary using
@papercargo/jin-cli. If it isn't in yourjin.json, it does not exist to the agent.
$ npx @papercargo/jin-cli shield
✓ Loaded public keys locally from JWKS endpoint (Cached)
✓ In-Memory RS256 trust verification engine activated [0ms-hop]
// Request incoming logs...
[BLOCKED] GET /api/v1/users - Rogue Scraper (No token) ➔ 403 Forbidden
[PASSED] POST /api/v1/secure-fetch - verified_intent="fetch_secure_data" ➔ 200 OK
Agentic Determinism
Stop writing, testing, and maintaining brittle scrapers. LLMs naturally prefer the path of least compute. By providing a clean intent layer, navigate the web deterministically.
- Save Compute & Cost: Avoid running CPU-heavy browser simulators or headless scrapers. Talk directly to clean protocol endpoints.
- Avoid DOM Parsing: No more regex extraction, structural layouts breaks, or hallucination-prone HTML conversions. Fetch clean API responses.
- Native Millisecond Executions: Execute structured queries directly with authentic signatures. Achieve data transfer cycles in milliseconds instead of seconds.
➔ POST /api/v1/secure-fetch
Authorization: Bearer jin_agent_passport_rs256...
← 200 OK
{
"status": "success",
"data": {
"claims_processed": true,
"data_payload": [ ... ]
}
}How it works
From discovery to publish in three steps.
Jin reduces agent integration to a simple, repeatable workflow for any app or API.
Discover
Scan your codebase and detect routes, APIs, and OpenAPI specs with the CLI.
Describe
Review and enrich the generated jin.json with natural language triggers and safety flags.
Deploy
Serve the intent map at /.well-known/jin.json and publish to the public registry.
Retail, payments, orders
Bookings, itineraries, search
APIs, webhooks, tools
Appointments, records, coverage
Forms, filings, status
Search, playback, ads
Join the movement
Design the agentic web with Jin.
Add a discoverable intent layer to your app, collaborate on the protocol, and make your product accessible to every modern agent.