Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/davidbuenov/dbv-specs-ops/llms.txt

Use this file to discover all available pages before exploring further.

Agent Readiness is a set of structured, standards-based files that make your web project or public API legible and discoverable by AI agents, crawlers, and LLM-powered tools. When activated, the SDD framework’s /spec, /build, and /ship phases automatically guide you through creating every required file — from a semantically annotated robots.txt to .well-known/ identity cards for your MCP server. The result is a project that cooperates cleanly with the growing ecosystem of AI-driven tooling, without any manual research on your part.
Agent Readiness is optional and only applies to projects with a public-facing web interface or API. If your project is a CLI tool, library, or private service, you can leave this feature disabled.

​
Activating Agent Readiness

To enable Agent Readiness for your project, set the following field in project.config.md: 
- **Agent Readiness (Web):** Yes
During bootstrap, the AI will detect a web or API stack and propose this flag automatically with an [ASSUMPTION: ...] marker. Once confirmed, the framework plans and implements all the files below during the /spec, /build, and /ship phases.

​
Implementation Checklist

The following nine concerns must be addressed to complete a full Agent Readiness integration. The /spec phase records them in docs/SPECIFICATIONS.md section 4.1; the /build phase implements them; and the /ship phase verifies that the correct HTTP Link headers are injected.
1

robots.txt — Crawler policy and Content-Signals

Create robots.txt at the project root. Include the Content-Signal header to declare your training and indexing policy, and point crawlers to your sitemap:
User-agent: *
Allow: /

# Content signals for AI agents and crawlers
Content-Signal: ai-train=no, search=yes, ai-input=yes

Sitemap: https://example.com/sitemap.xml
  • ai-train=no — content must not be used for model training.
  • search=yes — content may be indexed for semantic search.
  • ai-input=yes — agents may read and use content as context during inference.
2

/llms.txt — Semantic content map

Create /llms.txt at the root of your site. This Markdown file provides AI agents with a human-readable map of your content — what pages exist, what they contain, and how they relate to each other — so agents can navigate your project efficiently without parsing HTML.
# My Project — Content Map for AI

## What this project does
Brief description of the project purpose.

## Key pages
- [Home](/) — Overview and quick start
- [API Reference](/api) — Full endpoint documentation
- [Auth guide](/auth.md) — How bots authenticate

## Semantic sections
- /docs — User-facing documentation
- /api  — REST API endpoints (JSON)
3

/auth.md — Authentication guide for bots

Create /auth.md at the root. This Markdown file describes the registration, login, and OAuth flows that bots and agents should follow to access protected resources — separate from your human-facing documentation.
4

.well-known/api-catalog — RFC 9727 API catalog

Create .well-known/api-catalog following RFC 9727. This file lists your API endpoints and their link relations so that discovery tools can enumerate available services without manual configuration.
5

.well-known/ OAuth and signature metadata

Create the three OAuth and signature metadata files that enable secure agent-to-API communication:
  • .well-known/oauth-protected-resource — declares which resources are protected and which authorization server governs them.
  • .well-known/oauth-authorization-server — the standard OIDC/OAuth 2.0 server metadata document.
  • .well-known/http-message-signatures-directory — lists supported HTTP message signature algorithms for request verification.
6

.well-known/agent.json — Agent identity card

Create .well-known/agent.json to declare your project’s bot identity. This card tells other agents what your project’s agent is called, what it can do, and how to contact it:
{
  "name": "My Project Agent",
  "description": "Assists users with task X and Y.",
  "version": "1.0.0",
  "homepage": "https://example.com",
  "contact": "agent@example.com",
  "capabilities": [
    "answer-questions",
    "search-content"
  ],
  "auth": {
    "type": "oauth2",
    "authorization_url": "https://example.com/.well-known/oauth-authorization-server"
  }
}
7

.well-known/mcp.json — MCP server connection card

Create .well-known/mcp.json to advertise your Model Context Protocol server endpoint. This allows MCP-compatible agents and orchestrators to discover and connect to your server automatically:
{
  "mcp_version": "1.0",
  "server_name": "my-project-mcp",
  "server_url": "https://example.com/mcp",
  "transport": "http",
  "description": "MCP server exposing project skills and data.",
  "skills_index": "/.well-known/agent-skills/index.json"
}
8

.well-known/agent-skills/ — Skills index and manifests

Create the skills directory with two types of files:index.json — a machine-readable index of all skills your agent exposes:
{
  "skills": [
    {
      "id": "search",
      "name": "Content Search",
      "manifest": "/.well-known/agent-skills/search/SKILL.md"
    },
    {
      "id": "summarize",
      "name": "Page Summarizer",
      "manifest": "/.well-known/agent-skills/summarize/SKILL.md"
    }
  ]
}
SKILL.md manifests — one per skill, describing inputs, outputs, and usage examples in plain Markdown so both humans and agents can understand the skill’s contract.
9

Markdown content negotiation

Configure your hosting provider to serve the .md version of any page when an HTTP client sends the Accept: text/markdown header. This allows AI agents to receive clean, token-efficient Markdown instead of HTML.Add the appropriate Link response headers to your deployment configuration so that agents can discover the Markdown alternative, the MCP server card, and the API catalog from any page:Firebase (firebase.json):
{
  "hosting": {
    "headers": [
      {
        "source": "**",
        "headers": [
          {
            "key": "Link",
            "value": "</.well-known/agent-skills/index.json>; rel=\"agent-skills\", </.well-known/mcp.json>; rel=\"mcp-server-card\", </.well-known/api-catalog>; rel=\"api-catalog\""
          }
        ]
      }
    ]
  }
}
Netlify (netlify.toml):
[[headers]]
  for = "/*"
  [headers.values]
    Link = '''</.well-known/agent-skills/index.json>; rel="agent-skills", </.well-known/mcp.json>; rel="mcp-server-card", </.well-known/api-catalog>; rel="api-catalog"'''
The /ship phase’s Agent Readiness Verification step checks that these Link headers are present before the task is marked as done.

​
How it fits in the SDD cycle

PhaseAgent Readiness activity
/specEvaluates whether Agent Readiness (Web): Yes is set; adds section 4.1 checklist to docs/SPECIFICATIONS.md
/buildImplements all files above incrementally
/shipVerifies Link headers are injected

Build docs developers (and LLMs) love

Get started for freeTalk to us