Documentation
AIAGENTNET Docs
For builders and operators: AGIN, agent-native web, AHCP, AGPK, MCP market, downloads, and public API on https://www.aiagentnet.cloud. Topics are multi-page (left nav); use in-page anchors for subsections. Machine JSON lives under each chapter’s contract downloads.
Browse topics
What is AGIN
AGIN is an agent's identity card. In AIAGENTNET, AI employees are first-class principals—not anonymous bots. AGIN is how we name and recognize which agent is acting, the same way a human account names which person is signed in.
Why it exists
| Side | What AGIN gives them |
|---|---|
| Service providers (MCP builders, APIs, sites) | Know who used your service—which agent, under which org or owner context—not a faceless scraper farm. You can allow, rate-limit, or audit by accountable identity instead of only by IP or User-Agent. |
| Service users (people and organizations) | Know which agent is using a service on your behalf—core agent vs a task thread, sales bot vs support bot—so permission and review stay meaningful. |
A product goal: less "anti-bot" theater, more accountable automation
Today much of the public internet treats automated clients as hostile by default: CAPTCHAs, aggressive bot walls, IP bans. That is rational when callers are anonymous. AIAGENTNET's goal is to push wide AGIN adoption so legitimate agents present a stable, reviewable identity. Providers can prefer AGIN-bearing callers and reduce blunt anti-crawler friction for real work—not by hiding bots, but by making them identifiable and accountable.
AGIN is a signature AIAGENTNET capability. MCP on AIAGENTNET is not only "paste a tool into a desktop client." It is tools plus who is calling (AGIN) plus what happened (receipts), for both builders and users. See the MCP market.
What we publish publicly (builders & providers)
- Meaning: AGIN = agent identity card; who called / which agent is acting
- Call presentation: AGIN id, agent principal, call id (headers and/or
tools/call_meta.agentnet) - Separation: AGIN (which agent) vs third-party API keys (which product account)
- Receipts: log AGIN + tool + outcome for review
What we do not publish on the public docs site
- Key derivation paths, signing key ids, or platform crypto design
- HMAC secrets, private keys, or "how to mint AGIN yourself"
- Internal control-plane tables, gateway paths, or deployment topology
- Anything that would let a third party forge an agent identity
If a field is not in the public examples, treat it as not yet a stable public contract—or as internal-only.
Next steps
- Read What is AGIN, then the agent-native web paradigm
- Read AHCP 1.0 (Chrome sample + protocol JSON)
- Then Build or Use market connectors
- Download Server / Desktop ISO and verify checksums; follow the ISO deploy guide
- Create a free account for workspace and billing when you need them
- Pricing for paid office suite and Mail
- Status for live probes