Agntbase Connect site
Standard. Registry. Protocol.

Make your business readable for AI agents

Websites were built for people. Agntbase adds a minimal site manifest, a canonical entity in the hub, explicit access rules, and a structured protocol layer for agentic discovery and selection.

In one day: starter JSON and TXT, canonical hub profile, permissions manifest, install guidance, and verification link.
Core stack plus access rules
site manifest
hub profile
protocol layer
agent logs
Entrypoint JSON / TXT
Canonical Hub entity
Access rules Permissions
Access API / MCP-ready
Memory Logs / outcomes
Architecture

The stack is more than a file and a hub

A useful agent ecosystem needs an entrypoint on the client site, a canonical entity in the hub, explicit permissions, a structured protocol layer, and a logging layer for outcomes.

01

Entrypoint file

Short `json/txt` on the client site. Minimal pointer, not a full profile.

02

Canonical profile

Normalized company entity in the hub with attributes, scenarios, and limits.

03

Permissions layer

Explicit access rules for what agents may read, where canonical data lives, and which actions are off-limits.

04

Protocol layer

Structured access for agents through canonical JSON, API, and MCP-ready routes.

05

Analytics and logs

Requests, shortlist generation, profile opens, transitions, and outcome signals.

What you get fast

What a company gets in one working day

  • Minimal entrypoint JSON and TXT ready for placement on the site
  • Canonical company entity in the Agntbase hub
  • Permissions manifest that tells agents where to go next
  • Protocol URLs for machine access instead of HTML scraping
  • Verification link and install guidance for the website team
Packaging model

Free or low-touch setup starts with the entrypoint package, canonical profile, and install guide. Paid expansion adds manual review, clearer recommendation logic, and stronger verified claims.

Trust and verification

Profiles need provenance, not just polished copy

A useful hub profile should say who submitted it, when it was updated, what is self-attested, and where the canonical machine source lives.

  • Verification status for every canonical profile
  • Source-of-truth links and provenance records
  • Clear separation between starter package and reviewed profile
Why this matters

Without trust metadata, a registry becomes a prettier directory. With provenance and review paths, it becomes infrastructure that agents can reason about more safely.

Protocol layer

Agents should get structure, rules, and endpoints

Agntbase exposes a machine path that starts at the site manifest and continues through canonical JSON, permissions, search, and event logging.

Read

Canonical profile API

Profile JSON is available directly through a stable API URL for machine use.

Decide

Permissions manifest

Agents can read explicit access rules before deeper retrieval or action.

Log

Events and memory

Basic event ingestion and summaries create a path toward shortlist and outcome analytics.

Setup flow

How a company gets connected

Step 1

Submit the website

We review the site and prepare a minimal entrypoint and an initial canonical entity.

Step 2

Create the hub profile and rules

The company gets a canonical profile, permissions manifest, and clear routing to the hub JSON.

Step 3

Attach the protocol layer

Agents can use the canonical JSON, API endpoints, search, and later MCP-ready access.

What companies get

A cleaner business identity for AI-mediated selection

  • Machine-readable entrypoint on the client site
  • Canonical profile in a normalized hub
  • Permissions manifest for access clarity
  • Protocol-ready access for agents
  • Clear routing from file to canonical JSON
  • Event logging for future analytics
Why this matters

AI systems do not need more marketing pages. They need a clean entity, explicit access rules, and a canonical machine-readable source.

Install and verify

Placement guidance and public verification

The preferred installation path is /.well-known/agenthub.json. If that path is not available on the client site, publish the file as /agenthub.json. The same rule applies to the TXT entrypoint.

  • Preferred JSON: /.well-known/agenthub.json
  • Fallback JSON: /agenthub.json
  • Preferred TXT: /.well-known/agenthub.txt
  • Fallback TXT: /agenthub.txt

Verify placement

Enter a public website URL and Agntbase will check the expected entrypoint paths.

FAQ

Questions we expect early

Is this just another llms.txt generator?

No. The entrypoint file is only the first layer. The actual value is the canonical profile, structured access, and agent analytics.

Do companies need to understand the format themselves?

No. We prepare the draft package, routing, and placement guidance.

Why not just publish llms.txt on the website?

Because a file alone does not provide canonical structure, permissions, provenance, or protocol endpoints. Agntbase packages all of them together.

Will agents read the file automatically?

Not always. That is why the product is built around a registry and protocol layer, not only around a file on the website.

Can this evolve into API and MCP access?

Yes. The protocol layer is part of the product model from day one.

What is free and what becomes premium?

The starter package, canonical profile, and install flow can be delivered quickly. Manual review, clearer recommendation rules, and stronger verified claims are the premium layer.

Start

Connect your website

Leave a website URL and a contact. The starter package is created immediately, and the same screen will show the files, install paths, and verification link without waiting for email.