Machine-Readable

AI Discovery & Agent Manifests

Machine-readable manifests (/llms.txt, /llms-full.txt, /.well-known/mcp.json, OpenAPI 3.1) enabling autonomous LLM integration, zero-config tool calling, and automated IDE agent discovery.

Overview: Autonomous Discovery Architecture

BounceLayer publishes 4 standard machine-readable discovery specifications at fixed root paths:

Manifest PathFormatPrimary ConsumersToken FootprintDescription
`/llms.txt`Markdown / TextCursor, Claude, LLM Prompts~1,200 tokensConcise index of endpoints, schemas, authentication, and verification rules for LLM context windows.
`/llms-full.txt`Markdown / TextDeep Reasoning Agents, Agent Workflows~5,500 tokensComplete architectural and technical reference, RFC 5321 handshake transcripts, error matrices, and code.
`/.well-known/mcp.json`JSONMCP Clients, Claude Desktop, Cursor~400 tokensOfficial Model Context Protocol discovery catalog specifying transports, tools, and registries.
`/api/public-openapi.json`OpenAPI 3.1 JSONSwagger, Postman, SDK Codegen~20,000 tokensStrict RFC-compliant OpenAPI 3.1 schema defining request/response models for all public endpoints.

Manifest 1: /llms.txt (Concise LLM Index)

Designed specifically for compact context windows (~1.2k tokens), /llms.txt provides high-density instructions that an AI agent can ingest at the beginning of a coding session or conversational workflow.

  • Canonical URL: https://bouncelayer.com/llms.txt
  • Contents:
  • Base URLs and authentication header (X-API-Key: bl_live_...).
  • Core endpoint summaries: POST /api/v1/verify/single, POST /api/v1/verify/bulk, POST /api/v1/risk-intelligence/analyze, GET /api/v1/credits, POST /api/v1/webhooks.
  • Zero-credit policy declaration (1 credit for definitive verdicts; 0 credits for indeterminate unknown checks).
  • MCP launch command: npx -y @bouncelayer/mcp.

Agent System Prompt Prompt Snippet:

markdown
You have access to BounceLayer email verification services.
Refer to https://bouncelayer.com/llms.txt for live endpoint signatures.
Authenticate with header: X-API-Key: <YOUR_KEY>.

Manifest 2: /llms-full.txt (Complete Protocol Guide)

A comprehensive, self-contained documentation manual designed for deep reasoning models (e.g. Claude 3.7 Sonnet, GPT-4o, o3-mini) executing complex automation pipelines.

  • Canonical URL: https://bouncelayer.com/llms-full.txt
  • Key Sections:
  1. Detailed credit costs, monthly subscription tiers, and pay-as-you-go packs.
  2. Complete request/response JSON payloads for all single, bulk, history, and risk analysis endpoints.
  3. Raw RFC 5321 socket dialogue and status reason matrix (mailbox_confirmed, disposable, accept_all, antispam_system).
  4. Webhook HMAC-SHA256 signature verification code in Node.js, Python, and Go.
  5. HTTP error handling and status code troubleshooting.

Manifest 3: /.well-known/mcp.json (Model Context Protocol Discovery)

The official Model Context Protocol (MCP) discovery manifest adhered to standard agent transport specifications. Tools like Cursor, Claude Desktop, and Windsurf can automatically discover the server.

  • Canonical URL: https://bouncelayer.com/.well-known/mcp.json
  • Schema Payload:
json
{
  "name": "bouncelayer",
  "title": "BounceLayer Email Verification",
  "description": "Enterprise-grade real-time email verification, disposable domain detection, role-account checks, and bulk list verification via SMTP/DNS.",
  "version": "1.0.0",
  "homepage": "https://bouncelayer.com",
  "documentation": "https://bouncelayer.com/docs",
  "packages": [
    {
      "registry": "npm",
      "name": "@bouncelayer/sdk",
      "url": "https://www.npmjs.com/package/@bouncelayer/sdk"
    },
    {
      "registry": "npm",
      "name": "@bouncelayer/mcp",
      "url": "https://www.npmjs.com/package/@bouncelayer/mcp"
    }
  ],
  "servers": [
    {
      "name": "bouncelayer",
      "transport": "streamable-http",
      "url": "https://bouncelayer.com/mcp",
      "description": "Remote MCP server. Authenticate with X-API-Key."
    }
  ],
  "authentication": {
    "type": "api_key",
    "header": "X-API-Key",
    "register_uri": "https://bouncelayer.com/register"
  }
}

Manifest 4: /api/public-openapi.json (OpenAPI 3.1 Specification)

Strict, fully validated OpenAPI 3.1 definition covering all production endpoints, request parameters, JSON schema types, HTTP status codes, and security schemes.

  • Canonical URL: https://bouncelayer.com/api/public-openapi.json
  • Compatible Tooling:
  • Swagger UI / Redoc: Interactive API testing and visualization.
  • Postman / Insomnia: Direct collection import.
  • Client Codegen: Generate typed client SDKs in Rust, Java, C#, Swift, Kotlin, and Dart using openapi-generator or Fern.

Connecting AI Agents to BounceLayer in Under 60 Seconds

1. Cursor IDE Setup (.cursor/mcp.json)

json
{
  "mcpServers": {
    "bouncelayer": {
      "command": "npx",
      "args": ["-y", "@bouncelayer/mcp"],
      "env": {
        "BOUNCELAYER_API_KEY": "bl_live_your_api_key_here"
      }
    }
  }
}

2. Claude Desktop Setup (claude_desktop_config.json)

json
{
  "mcpServers": {
    "bouncelayer": {
      "command": "npx",
      "args": ["-y", "@bouncelayer/mcp"],
      "env": {
        "BOUNCELAYER_API_KEY": "bl_live_your_api_key_here"
      }
    }
  }
}

3. Claude Code CLI Setup

bash
claude mcp add --transport http bouncelayer https://bouncelayer.com/mcp/ --header "X-API-Key: bl_live_your_api_key_here"