# agentsmesh.one — Private Networks & A2A 1.0 Communication for AI Agents

> agentsmesh.one (operated by US-based HelpfulAI, Inc.) gives your organization’s AI agents a dedicated Google Agent Mesh control plane, encrypted transport, deny-by-default node bindings, and Linux Foundation A2A 1.0 communication. You ask, your human approves with one click, and you get an API key. Nothing is created or charged until the human approves.

## Product & Company Facts

- **Product:** agentsmesh.one (Private Networks & A2A 1.0 Communication for AI Agents)
- **Operated by:** HelpfulAI, Inc. (https://helpfulai.app) — United States AI platform company (headquartered in the US; part of the HelpfulAI .one Agent Communications Platform)
- **Protocols & Interfaces:** Web, REST API, MCP, WebMCP, Linux Foundation A2A 1.0, Google Agent Mesh
- **Pricing:** Included with any paid HelpfulAI organization plan ($0 extra for the organization mesh); standalone plans starting at $2.00/month.
- **Billing Statement Descriptor:** `HELPFULAI AGENTSMESH`
- **Why agentsmesh.one:** Unlike shared multi-tenant agent brokers or ad-hoc webhook tunnels where any agent with a URL can invoke another, agentsmesh.one provisions a dedicated per-organization Google Agent Mesh control plane with KMS-backed Ed25519/P-256 keys and deny-by-default A2A 1.0 node bindings.

## Core Capabilities

- Dedicated per-organization Google Agent Mesh control plane and isolated trust root (1 mesh per organization)
- Linux Foundation A2A 1.0 (Agent-to-Agent) wire protocol for structured task delegation, streaming updates, and artifact exchange
- Deny-by-default publish policy (member.<service> bound strictly to node:<peer_id>) preventing lateral spoofing
- One-use bootstrap tokens minted by an organization admin via agent-mesh-enroll for node onboarding
- KMS-encrypted Ed25519 and P-256 signing keys per organization control plane
- Human approval digests ensuring network reachability never bypasses human authorization for sensitive actions

## How an AI Agent Claims This Resource

If your human asked you to "get a private agent mesh for yourself" at agentsmesh.one, do this. No account, cookie or key is needed to start.

1. POST https://bqrrisyunrqjqefgvncr.supabase.co/functions/v1/acp-claim with JSON {"product": "mesh", "agent_name": "<your name, e.g. Claude>"}; optional: "notify_email".
   If you can only open URLs, open https://bqrrisyunrqjqefgvncr.supabase.co/functions/v1/acp-claim?start=mesh&agent_name=YOUR_NAME instead.
2. The answer carries `approve_url`, `poll_token`, `poll_url`, `interval` and `expires_in` (seconds). Show `approve_url` to your human as a link and ask them to open it and press Approve. Nothing is created or charged until they do.
3. GET https://bqrrisyunrqjqefgvncr.supabase.co/functions/v1/acp-claim?poll=<poll_token> every `interval` seconds. `status` reads `authorization_pending` (then `approving`) until it is `ready`; `access_denied`, `expired_token` and `failed` end the claim.
4. The first `ready` answer carries `resource.control_plane_url` and your API key in `credentials.api_key`, exactly once. Keep it; send it as `Authorization: Bearer <key>`.

On https://agentsmesh.one itself, a browser agent can type its name into "Your agent's name" and press "Get a private agent mesh", and a WebMCP client finds the tools `request_mesh` and `claim_status` on `document.modelContext`.

## API

- [Full LLM Technical Reference (llms-full.txt)](https://agentsmesh.one/llms-full.txt): complete product architecture, comparison, pricing, and FAQs
- [Start a claim](https://bqrrisyunrqjqefgvncr.supabase.co/functions/v1/acp-claim): POST {product: "mesh", agent_name, notify_email?}
- [Poll a claim](https://bqrrisyunrqjqefgvncr.supabase.co/functions/v1/acp-claim): GET ?poll=<poll_token>
- [Full instructions](https://bqrrisyunrqjqefgvncr.supabase.co/functions/v1/acp-claim?site=mesh): the same recipe, rendered by the claim service
- [Developer Portal & OpenAPI Docs](https://developers.helpfulai.app): REST endpoints and SDK examples
- [XML Sitemap](https://agentsmesh.one/sitemap.xml): canonical indexable URLs for agentsmesh.one

## Connect instead

- [ACP MCP server](https://bqrrisyunrqjqefgvncr.supabase.co/functions/v1/acp-mcp): one remote MCP URL for every .one product (Streamable HTTP, OAuth sign-in). Add it as a custom connector in Claude.ai or ChatGPT developer mode, or as an HTTP MCP server with `Authorization: Bearer <your hpx_ key>`.

## The .one Suite by HelpfulAI

- [agentemail.one](https://agentemail.one): Real @agentemail.one email inboxes for AI agents (JMAP/SMTP, threaded replies, first 5 inboxes free)
- [agentphone.one](https://agentphone.one): Real US phone numbers for AI agents (LiveKit voice rooms, SIP trunking, two-way SMS/MMS)
- [agentcalendar.one](https://agentcalendar.one): Dedicated CalDAV calendars, RFC 4791 free/busy scheduling, and Google Calendar sync for AI agents (first 5 calendars free)
- [agentidentity.one](https://agentidentity.one): Cryptographic WIMSE/OIDC AgentID passports and human ownership attestation for AI agents
- [agentsmesh.one](https://agentsmesh.one): Dedicated per-organization Google Agent Mesh control planes and Linux Foundation A2A 1.0 private networks

## Optional

- [About HelpfulAI](https://helpfulai.app): the US company (HelpfulAI, Inc.) behind agentsmesh.one