Skip to content
v1.0 · Open specification

The Bippsi AI Standard

One unified manifest per site. Agents fetch /bippsi-unified.md once and get everything — identity, policy, MCP endpoint, API, certification, page index. No more ten-file hunts.

Public repo: github.com/bigappstudiollc/bippsi

What we mean by "Standard": The "Bippsi AI Standard" is an open specification authored and maintained by Bippsi (Big App Studio LLC). It is not a standard issued by, endorsed by, or adopted by IETF, W3C, ISO, or any independent standards-development organization. We publish the spec under a permissive license so anyone can implement it; adoption is voluntary. Compliance with this specification is not a warranty of site quality, security, or agent safety.

Why does this exist?

Today, an AI agent visiting a new domain has to fetch up to ten separate files just to learn what it can do there:

robots.txt
sitemap.xml
llms.txt
AGENTS.md
agents.json
openapi.json
ai-plugin.json
mcp.json
security.txt
manifest.json

Ten round-trips. Ten parse paths. Ten places something can drift. The Bippsi AI Standard replaces this with one file at one known location.

Layer 1 · Legacy

The broken system

Every site should still publish the individual discovery files. Older agents depend on them, and they document specific sub-concerns (robots, security disclosure, MCP) cleanly.

  • 10+ files across different paths
  • Multiple round-trips to piece together a full picture
  • Drift-prone — files can disagree with each other
  • Works today — the foundation of every agent-aware web tool
Layer 2 · Unified

The unified system

A single markdown file at /bippsi-unified.md with 9 required sections covering everything an agent needs. One file, one round-trip, authoritative.

  • 1 file at a known root path
  • Human + machine readable — markdown with fenced JSON blocks
  • Authoritative — if it disagrees with legacy files, it wins
  • Auto-generated when produced by Bippsi plugins or code

Agent discovery algorithm

A conforming Bippsi-aware agent should follow this order when first encountering a domain:

# 1. Canonical entry point — one IETF well-known URL lists every surface
GET /.well-known/bippsi.json

# 2. Unified manifest — full machine-readable policy + pricing + page index
GET /bippsi-unified.md

# 3. MCP manifest (if the site runs an MCP server)
GET /.well-known/mcp.json

# 4. Legacy individual files (parallel fetch)
GET /AGENTS.md
GET /llms.txt
GET /openapi.json
GET /agents.json

# 5. Last resort: HTML scraping + schema.org JSON-LD extraction

Time budget: step 1 is a single round-trip and tells the agent which subsequent fetches are worthwhile. Step 2 collapses identity, policy, pricing, and page index into one request. Steps 3-4 are only needed when the site hasn't adopted the unified manifest yet. Step 5 is the traditional scraping path.

Canonical entry point

/.well-known/bippsi.json

A single IETF RFC 8615 well-known URL that lists every AI-discovery artifact the site publishes — the unified manifest, the legacy llms.txt / AGENTS.md, the MCP manifest, the OpenAPI spec, the optional /bippsi/search / /bippsi/topics / /bippsi/headlines endpoints, and protocol contact URLs. Stable shape with nulls where a surface isn't implemented yet, so agents never need to probe.

Auto-emitted by the Bippsi WordPress plugin and General handler on every certified Partner site.

Required sections of /bippsi-unified.md

#SectionContents
1IdentityName, url, tagline, contact, same_as (as JSON)
2Agent access policyWhat agents may/may not do, auth methods, rate limits
3MCP serverEndpoint, protocol version, tool list
4REST APIBase URL, OpenAPI link, endpoint list (as JSON)
5Certification statusA.I. Certified score, verify URL
6Content licensingAI training, reproduction, redistribution terms
7Page indexPublic pages list (equivalent to sitemap + llms.txt)
8Fallback filesPointers to the legacy individual files
9Spec metadataSpec version, generation timestamp, publisher

Optional sections may follow: Agent payments (10), Capabilities (11), Changelog (12).

MCP discovery convention

The Model Context Protocol spec standardizes the protocol itself but does not define a remote-server discovery URL. Until it does, this standard proposes:

A site running an MCP server SHOULD publish a JSON manifest at /.well-known/mcp.json describing its endpoint, protocol version, capabilities, and tools. Agents SHOULD check this path first before attempting legacy /mcp or subdomain-based discovery.

Bippsi's own MCP server at https://bippsi.com/api/v1/mcp is the reference implementation; its manifest at /.well-known/mcp.json is the canonical example.

See it in action

Bippsi.com itself publishes a conforming unified manifest. Fetch it, parse it, or just read it:

How to adopt

WordPress customers: The Bippsi WordPress plugin v3.3+ will emit /bippsi-unified.md automatically. No action needed beyond upgrading the plugin when it ships.
General Website customers: Same — the General Website plugin v2.1+ will generate the manifest from your deployed config. Automatic.
Custom stacks: Drop in one of the portable reference implementations from the public repo (PHP, Node, or Python — all dependency-free, single file) or hand-author. Spec requires the 9 sections listed above with valid JSON blocks. Live example: Bippsi's own /bippsi-unified.md.
No website, just building an agent? Use the discovery algorithm above. When a site publishes a unified manifest, prefer it. Fall back gracefully for sites that haven't adopted yet.

Versioning and contact

This standard uses semver. Breaking changes bump major; additive changes bump minor. Current version: 1.0 (2026-04-14).

Proposals and bugs: open an issue at github.com/bigappstudiollc/bippsi/issues. Private or commercial questions: use the contact form.

What is Bippsi?

Bippsi is the agent-native layer of the web — a suite of apps and a platform that gives AI agents identity, payment, and compliant access to websites. Formerly Big App Studio.

How does Agent Initiative certify a website?

The scanner tests 15 compliance categories and 100+ checks — from structured data and llms.txt discovery through security headers and agent-native payment declarations. Sites scoring 85% or higher receive a public A.I. Certified badge.

Where can AI agents find Bippsi's access policy?

Everything live for agents is at /AGENTS.md, /llms.txt, /agents.json, and /openapi.json.

API endpoint: /api/v1/validate · OpenAPI: /openapi.json · MCP: /api/v1/mcp · Unified manifest: /bippsi-unified.md