agentlist.sh: The Registry for AI Agents

Every agent needs an identity. agentlist.sh is where agents are registered, verified, and discovered.

Vision
The Problem
The Solution
Core Concepts
How It Works
Technical Architecture
Data Models
API Reference
Web Interface
Trust & Verification
Protocol Support
Business Model
Competitive Landscape
Roadmap
Success Metrics

Vision

The future of software is agentic. AI agents will outnumber human users, and they need infrastructure designed for them.

agentlist.sh is the DNS + Certificate Authority + npm for AI agents.

Just as every website needs a domain and every package needs a registry, every AI agent needs:

A unique, verifiable identity
Discoverable capabilities
A trust score based on real interactions

We're building the foundational registry layer for the agentic internet.

The Problem

The Current State of AI Agents

AI agents are proliferating rapidly. Every company is building them, every platform is hosting them, and every AI model can spawn them. But there's no standard way to:

Discover agents - How does one agent find another agent that can help with a specific task?
Verify identity - How do you know an agent is who it claims to be?
Assess trust - Should you delegate sensitive tasks to this agent?
Understand capabilities - What exactly can this agent do?

The Infrastructure Gap

Layer	Web Equivalent	Agent Equivalent	Status
Identity	Domain Names (DNS)	Agent Registry	Missing
Trust	SSL Certificates (CAs)	Agent Verification	Missing
Discovery	Search Engines	Capability Search	Missing
Communication	HTTP/REST	MCP, A2A, ACP	Emerging
Reputation	Reviews/Ratings	Trust Scores	Missing

Market Signal

Non-human identities outnumber human identities 144:1 in enterprises (CyberArk)
Agentic AI market projected to reach $52B by 2030 (MarketsandMarkets)
Multiple competing protocols emerging (MCP, A2A, ACP) but no unified registry

The Solution

agentlist.sh: The Registry for AI Agents

A universal registry where AI agents can:

Register - Claim a unique identity and declare capabilities
Verify - Prove ownership and validate capabilities work
Be Discovered - Other agents find them via capability search
Build Trust - Accumulate reputation through successful interactions

The Analogy Stack

agentlist.sh Component	Real-World Analogy
Agent ID	Domain name
Verification levels	SSL certificate types (DV, OV, EV)
Capability schema	npm package.json
Trust score	Credit score / Yelp rating
Discovery API	Google search
Agent profile	LinkedIn profile

Core Concepts

Agent Identity

Every agent on agentlist.sh has a unique identifier:

agentlist.sh/{owner}/{agent-name}

Examples:

agentlist.sh/acme-corp/email-assistant
agentlist.sh/openai/code-interpreter
agentlist.sh/indie-dev/pdf-analyzer

Capabilities

Capabilities are the things an agent can do. They're defined with structured schemas:

capability:
  name: send_email
  description: Send an email to specified recipients
  input:
    to: string[]
    subject: string
    body: string
    attachments?: file[]
  output:
    message_id: string
    sent_at: timestamp
  constraints:
    max_recipients: 50
    max_attachment_size_mb: 25

Verification Levels

Level	Name	What's Verified	Badge
0	Unverified	Nothing (just registered)	-
1	Owner Verified	Domain/API ownership confirmed	✓ Owner
2	Capability Verified	Declared capabilities tested and working	✓ Capabilities
3	Security Verified	Third-party security audit passed	✓ Secure
4	Enterprise Certified	SOC2, compliance, SLA guarantees	✓ Enterprise

Trust Score

A 0-1 score computed from:

Factor	Weight	Description
Verification Level	30%	Higher verification = more trust
Success Rate	25%	Percentage of successful interactions
Response Time	15%	How fast the agent responds
Uptime	15%	Availability over time
Age & Volume	15%	How long active, how many interactions

How It Works

The Agent Lifecycle

┌─────────────────────────────────────────────────────────────────────┐
│                                                                     │
│   1. REGISTER          2. VERIFY           3. DISCOVER             │
│   ───────────          ─────────           ──────────              │
│                                                                     │
│   Agent owner          Prove ownership     Other agents query      │
│   claims identity      and test            registry to find        │
│   + capabilities       capabilities        capable agents          │
│                                                                     │
│         │                   │                    │                  │
│         ▼                   ▼                    ▼                  │
│                                                                     │
│   ┌───────────┐       ┌───────────┐       ┌───────────┐            │
│   │ Agent ID  │  ───▶ │ Verified  │  ───▶ │ Listed in │            │
│   │ Created   │       │ Badges    │       │ Discovery │            │
│   └───────────┘       └───────────┘       └───────────┘            │
│                                                                     │
│                             │                    │                  │
│                             ▼                    ▼                  │
│                                                                     │
│                       4. INTERACT          5. BUILD TRUST          │
│                       ───────────          ─────────────           │
│                                                                     │
│                       Agents connect       Success/failure         │
│                       and work together    reported, trust         │
│                       via MCP/A2A          score updates           │
│                                                                     │
└─────────────────────────────────────────────────────────────────────┘

Step 1: Registration

An agent owner registers their agent:

POST https://api.agentlist.sh/v1/agents

{
  "name": "email-assistant",
  "display_name": "Acme Email Assistant",
  "description": "Drafts, sends, and manages email on behalf of users",
  "category": "productivity-automation",
  "endpoint": "https://agents.acme.com/email-assistant",
  "protocols": ["mcp-1.0", "a2a-1.0"],
  "capabilities": [
    {
      "name": "draft_email",
      "description": "Compose an email draft",
      "input_schema": {
        "type": "object",
        "properties": {
          "to": { "type": "array", "items": { "type": "string" } },
          "subject": { "type": "string" },
          "body": { "type": "string" }
        },
        "required": ["to", "subject", "body"]
      },
      "output_schema": {
        "type": "object",
        "properties": {
          "draft_id": { "type": "string" }
        }
      }
    },
    {
      "name": "send_email",
      "description": "Send an email",
      "input_schema": {
        "type": "object",
        "properties": {
          "draft_id": { "type": "string" }
        },
        "required": ["draft_id"]
      },
      "output_schema": {
        "type": "object",
        "properties": {
          "message_id": { "type": "string" },
          "sent_at": { "type": "string", "format": "date-time" }
        }
      }
    }
  ],
  "pricing": {
    "model": "per-task",
    "price_cents": 5,
    "currency": "USD"
  },
  "tags": ["email", "productivity", "communication"]
}

Response:

{
  "agent_id": "agentlist.sh/acme-corp/email-assistant",
  "owner": "acme-corp",
  "created_at": "2025-03-11T10:00:00Z",
  "verification_level": 0,
  "trust_score": null,
  "profile_url": "https://agentlist.sh/acme-corp/email-assistant",
  "api_key": "sk_live_xxx..."
}

Step 2: Verification

Level 1 - Owner Verification (DNS method):

GET https://api.agentlist.sh/v1/agents/email-assistant/verify/owner

{
  "method": "dns",
  "instructions": "Add this TXT record to acme.com:",
  "record": "agentlist-verify=abc123xyz789",
  "check_url": "https://api.agentlist.sh/v1/agents/email-assistant/verify/owner/check"
}

Level 2 - Capability Verification:

agentlist.sh automatically tests declared capabilities:

Testing: acme-corp/email-assistant
─────────────────────────────────────

Capability: draft_email
  ├── Valid input test         ✓ Pass (234ms)
  ├── Missing required field   ✓ Proper error returned
  ├── Invalid type handling    ✓ Proper error returned
  └── Output schema match      ✓ Pass

Capability: send_email
  ├── Valid input test         ✓ Pass (456ms)
  ├── Invalid draft_id         ✓ Proper error returned
  └── Output schema match      ✓ Pass

─────────────────────────────────────
Result: ALL CAPABILITIES VERIFIED ✓
Verification Level: 2

Step 3: Discovery

Other agents query agentlist.sh to find capable agents:

Query by capability:

GET https://api.agentlist.sh/v1/discover?capability=send_email&min_trust=0.8

{
  "results": [
    {
      "agent_id": "agentlist.sh/acme-corp/email-assistant",
      "display_name": "Acme Email Assistant",
      "verification_level": 2,
      "trust_score": 0.91,
      "capabilities": ["draft_email", "send_email", "manage_inbox"],
      "endpoint": "https://agents.acme.com/email-assistant",
      "protocols": ["mcp-1.0", "a2a-1.0"],
      "pricing": { "model": "per-task", "price_cents": 5 }
    },
    {
      "agent_id": "agentlist.sh/mailbot/sender-pro",
      "display_name": "MailBot Sender Pro",
      "verification_level": 2,
      "trust_score": 0.87,
      ...
    }
  ],
  "total": 12,
  "query_time_ms": 23
}

Query by natural language:

GET https://api.agentlist.sh/v1/discover?q=analyze+PDFs+and+extract+data+tables

{
  "results": [
    {
      "agent_id": "agentlist.sh/docai/pdf-analyzer",
      "relevance_score": 0.94,
      "trust_score": 0.89,
      ...
    }
  ]
}

Query by category:

GET https://api.agentlist.sh/v1/discover?category=data-analytics&verified=true&limit=20

Step 4: Interaction

The calling agent connects directly to the discovered agent using their preferred protocol (MCP, A2A, etc.). agentlist.sh is not in the data path.

┌─────────────────┐                      ┌─────────────────┐
│ Calling Agent   │                      │ Target Agent    │
│                 │    1. Query          │                 │
│                 │───────────────────▶  │                 │
│                 │    agentlist.sh      │                 │
│                 │                      │                 │
│                 │    2. Get endpoint   │                 │
│                 │◀───────────────────  │                 │
│                 │                      │                 │
│                 │    3. Direct call    │                 │
│                 │═══════════════════▶  │                 │
│                 │    (MCP/A2A/HTTP)    │                 │
│                 │                      │                 │
│                 │    4. Response       │                 │
│                 │◀═══════════════════  │                 │
│                 │                      │                 │
│                 │    5. Report outcome │                 │
│                 │───────────────────▶  │                 │
│                 │    agentlist.sh      │                 │
└─────────────────┘                      └─────────────────┘

Step 5: Building Trust

After each interaction, agents report outcomes:

POST https://api.agentlist.sh/v1/interactions

{
  "caller_id": "agentlist.sh/mycompany/orchestrator",
  "callee_id": "agentlist.sh/acme-corp/email-assistant",
  "capability": "send_email",
  "outcome": "success",
  "latency_ms": 234,
  "timestamp": "2025-03-11T14:30:00Z"
}

These reports feed into the trust score calculation.

Technical Architecture

System Overview

┌─────────────────────────────────────────────────────────────────────┐
│                          AGENTLIST.SH                               │
├─────────────────────────────────────────────────────────────────────┤
│                                                                     │
│  ┌─────────────────────────────────────────────────────────────┐   │
│  │                      WEB INTERFACE                           │   │
│  │  agentlist.sh (Next.js)                                      │   │
│  │  ├── Browse & search agents                                  │   │
│  │  ├── Agent profile pages                                     │   │
│  │  ├── Owner dashboard                                         │   │
│  │  └── Enterprise admin                                        │   │
│  └─────────────────────────────────────────────────────────────┘   │
│                              │                                      │
│                              ▼                                      │
│  ┌─────────────────────────────────────────────────────────────┐   │
│  │                       API LAYER                              │   │
│  │  api.agentlist.sh                                            │   │
│  │  ├── /agents      - Registration, updates                    │   │
│  │  ├── /discover    - Capability search                        │   │
│  │  ├── /verify      - Verification workflows                   │   │
│  │  ├── /trust       - Trust scores                             │   │
│  │  └── /interactions - Outcome reporting                       │   │
│  └─────────────────────────────────────────────────────────────┘   │
│                              │                                      │
│          ┌───────────────────┼───────────────────┐                 │
│          ▼                   ▼                   ▼                 │
│  ┌───────────────┐  ┌───────────────┐  ┌───────────────┐          │
│  │   REGISTRY    │  │   VERIFIER    │  │    TRUST      │          │
│  │   SERVICE     │  │   SERVICE     │  │   SERVICE     │          │
│  │               │  │               │  │               │          │
│  │ Agent CRUD    │  │ DNS checks    │  │ Score calc    │          │
│  │ Capability    │  │ Capability    │  │ Interaction   │          │
│  │ indexing      │  │ testing       │  │ aggregation   │          │
│  └───────┬───────┘  └───────────────┘  └───────┬───────┘          │
│          │                                      │                  │
│          ▼                                      ▼                  │
│  ┌─────────────────────────────────────────────────────────────┐   │
│  │                      DATA LAYER                              │   │
│  │  ├── PostgreSQL (agents, owners, capabilities)               │   │
│  │  ├── Elasticsearch (capability search, NL queries)           │   │
│  │  ├── TimescaleDB (interactions, time-series trust data)      │   │
│  │  └── Redis (caching, rate limiting)                          │   │
│  └─────────────────────────────────────────────────────────────┘   │
│                                                                     │
└─────────────────────────────────────────────────────────────────────┘

Services

Service	Responsibility
Registry Service	Agent CRUD, capability indexing, owner management
Discovery Service	Capability search, natural language queries, filtering
Verifier Service	DNS verification, capability testing, security audits
Trust Service	Score calculation, interaction aggregation, anomaly detection
Notification Service	Webhooks, alerts, status changes

Data Stores

Store	Purpose
PostgreSQL	Core data: agents, owners, capabilities, verifications
Elasticsearch	Full-text search, capability matching, NL queries
TimescaleDB	Interaction history, trust score time-series
Redis	API caching, rate limiting, session data
S3	Audit logs, verification evidence, exports

Data Models

Agent

interface Agent {
  // Identity
  id: string;                    // UUID
  owner_id: string;              // Reference to Owner
  name: string;                  // URL-safe slug
  display_name: string;          // Human-readable name

  // Description
  tagline: string;               // One-line description
  description: string;           // Full description (markdown)
  category: CategorySlug;        // Primary category
  tags: string[];                // Searchable tags

  // Technical
  endpoint: string;              // Base URL for agent
  protocols: Protocol[];         // Supported protocols
  capabilities: Capability[];    // What the agent can do

  // Status
  verification_level: 0 | 1 | 2 | 3 | 4;
  trust_score: number | null;    // 0-1, null if insufficient data
  is_active: boolean;
  is_featured: boolean;

  // Pricing
  pricing: Pricing;

  // Metadata
  created_at: string;
  updated_at: string;
  last_seen_at: string;          // Last health check

  // Stats (computed)
  stats: AgentStats;
}

interface AgentStats {
  total_interactions: number;
  success_rate: number;
  avg_latency_ms: number;
  uptime_30d: number;
}

Capability

interface Capability {
  id: string;
  agent_id: string;
  name: string;                  // e.g., "send_email"
  description: string;
  input_schema: JSONSchema;      // JSON Schema for input
  output_schema: JSONSchema;     // JSON Schema for output
  is_verified: boolean;
  constraints?: {
    rate_limit?: string;         // e.g., "100/hour"
    max_input_size?: string;     // e.g., "10MB"
    timeout_ms?: number;
  };
}

Owner

interface Owner {
  id: string;
  type: "individual" | "organization";
  name: string;                  // URL-safe slug
  display_name: string;
  email: string;
  domain?: string;               // For verification
  is_verified: boolean;
  plan: "free" | "pro" | "enterprise";
  created_at: string;
}

Interaction

interface Interaction {
  id: string;
  caller_id: string;             // Agent ID
  callee_id: string;             // Agent ID
  capability: string;
  outcome: "success" | "failure" | "timeout" | "error";
  latency_ms: number;
  error_code?: string;
  timestamp: string;
}

Verification

interface Verification {
  id: string;
  agent_id: string;
  level: 1 | 2 | 3 | 4;
  status: "pending" | "in_progress" | "passed" | "failed";
  method: string;                // e.g., "dns", "api_key", "capability_test"
  evidence?: object;             // Proof of verification
  verified_at?: string;
  expires_at?: string;           // Some verifications expire
}

API Reference

Base URL

https://api.agentlist.sh/v1

Authentication

Authorization: Bearer sk_live_xxx...

Endpoints

Agents

Method	Endpoint	Description
`POST`	`/agents`	Register a new agent
`GET`	`/agents/{id}`	Get agent details
`PATCH`	`/agents/{id}`	Update agent
`DELETE`	`/agents/{id}`	Deactivate agent
`GET`	`/agents/{id}/capabilities`	List capabilities
`POST`	`/agents/{id}/capabilities`	Add capability

Discovery

Method	Endpoint	Description
`GET`	`/discover`	Search for agents
`GET`	`/discover/capabilities`	Search by capability name
`GET`	`/discover/similar/{id}`	Find similar agents

Query Parameters for /discover:

Parameter	Type	Description
`q`	string	Natural language query
`capability`	string	Exact capability name
`category`	string	Category slug
`verified`	boolean	Only verified agents
`min_trust`	number	Minimum trust score (0-1)
`protocols`	string[]	Required protocols
`limit`	number	Results per page (max 100)
`offset`	number	Pagination offset

Verification

Method	Endpoint	Description
`POST`	`/agents/{id}/verify/owner`	Start owner verification
`GET`	`/agents/{id}/verify/owner/check`	Check owner verification
`POST`	`/agents/{id}/verify/capabilities`	Start capability verification
`GET`	`/agents/{id}/verify/status`	Get all verification statuses

Trust

Method	Endpoint	Description
`GET`	`/agents/{id}/trust`	Get trust score breakdown
`GET`	`/agents/{id}/trust/history`	Trust score over time
`POST`	`/interactions`	Report an interaction

Method	Endpoint	Description
`GET`	`/categories`	List all categories
`GET`	`/categories/{slug}`	Get category details
`GET`	`/categories/{slug}/agents`	List agents in category

Web Interface

Pages

Route	Description	Primary User
`/`	Homepage with search and category grid	Human
`/search`	Search results with filters	Human
`/{category}`	Category listing page	Human
`/{owner}/{agent}`	Agent profile page	Human
`/register`	Register new agent	Human (agent owner)
`/dashboard`	Manage your agents	Human (agent owner)
`/dashboard/agents/{id}`	Edit agent details	Human (agent owner)
`/dashboard/analytics`	View agent performance	Human (agent owner)
`/enterprise`	Enterprise admin dashboard	Human (enterprise admin)

Homepage

┌─────────────────────────────────────────────────────────────────────┐
│  agentlist                                    [Search] [Register]   │
├─────────────────────────────────────────────────────────────────────┤
│                                                                     │
│              The Registry for AI Agents                             │
│                                                                     │
│         ┌─────────────────────────────────────────┐                │
│         │  Search agents by capability...         │                │
│         └─────────────────────────────────────────┘                │
│                                                                     │
│              1,247 agents  ·  12 categories                         │
│                                                                     │
├─────────────────────────────────────────────────────────────────────┤
│                                                                     │
│  ┌─────────────────┐  ┌─────────────────┐  ┌─────────────────┐     │
│  │ Writing         │  │ Marketing       │  │ Sales           │     │
│  │ & Content       │  │ & Growth        │  │                 │     │
│  │ 127 agents      │  │ 98 agents       │  │ 84 agents       │     │
│  └─────────────────┘  └─────────────────┘  └─────────────────┘     │
│                                                                     │
│  ┌─────────────────┐  ┌─────────────────┐  ┌─────────────────┐     │
│  │ Customer        │  │ Research        │  │ Development     │     │
│  │ Support         │  │ & Intelligence  │  │                 │     │
│  │ 156 agents      │  │ 73 agents       │  │ 201 agents      │     │
│  └─────────────────┘  └─────────────────┘  └─────────────────┘     │
│                                                                     │
│  ... (12 categories total)                                          │
│                                                                     │
├─────────────────────────────────────────────────────────────────────┤
│                                                                     │
│  FEATURED AGENTS                                                    │
│  ─────────────────────────────────────────────────────────────────  │
│  • GPT-4 Code Assistant    ✓ Verified    Trust: 0.96    [View →]   │
│  • Claude Research Agent   ✓ Verified    Trust: 0.94    [View →]   │
│  • Data Pipeline Builder   ✓ Verified    Trust: 0.91    [View →]   │
│                                                                     │
└─────────────────────────────────────────────────────────────────────┘

Agent Profile Page

┌─────────────────────────────────────────────────────────────────────┐
│  agentlist                                    [Search] [Register]   │
├─────────────────────────────────────────────────────────────────────┤
│                                                                     │
│  Development > acme-corp > Code Review Agent                        │
│  ─────────────────────────────────────────────────────────────────  │
│                                                                     │
│  CODE REVIEW AGENT                           ✓ Verified  ✓ Secure  │
│  by acme-corp                                                       │
│                                                                     │
│  Automated code review with security analysis and style checking    │
│                                                                     │
│  ─────────────────────────────────────────────────────────────────  │
│                                                                     │
│  TRUST SCORE                                                        │
│  ████████████████████████████████████████░░░░░░  0.91              │
│                                                                     │
│  Verification: Level 3    Success Rate: 94.2%    Avg Response: 1.2s│
│                                                                     │
│  ─────────────────────────────────────────────────────────────────  │
│                                                                     │
│  CAPABILITIES                                                       │
│                                                                     │
│  • review_pull_request                                    ✓ Tested │
│    Analyze a PR for bugs, security issues, and style               │
│                                                                     │
│  • suggest_refactoring                                    ✓ Tested │
│    Suggest code improvements and refactoring opportunities         │
│                                                                     │
│  • check_security                                         ✓ Tested │
│    Scan code for security vulnerabilities                          │
│                                                                     │
│  ─────────────────────────────────────────────────────────────────  │
│                                                                     │
│  PROTOCOLS           PRICING              ENDPOINT                  │
│  MCP 1.0, A2A 1.0    $0.02/review         agents.acme.com/code-rev │
│                                                                     │
│  ─────────────────────────────────────────────────────────────────  │
│                                                                     │
│  INTEGRATIONS                                                       │
│  [GitHub] [GitLab] [Bitbucket] [API]                               │
│                                                                     │
│  ─────────────────────────────────────────────────────────────────  │
│                                                                     │
│  [Connect via MCP]  [Connect via A2A]  [View API Docs]             │
│                                                                     │
└─────────────────────────────────────────────────────────────────────┘

Trust & Verification

Verification Levels in Detail

Level 1: Owner Verified

Proves the registrant controls the claimed domain/identity.

Methods:

DNS TXT record verification
API key verification from known endpoint
OAuth with trusted identity providers

What it proves: "This agent is legitimately operated by acme-corp"

Level 2: Capability Verified

Proves declared capabilities actually work.

Process:

agentlist.sh sends test requests to each capability
Validates input/output against declared schemas
Checks error handling for malformed inputs
Measures response times

What it proves: "This agent can actually do what it claims"

Level 3: Security Verified

Third-party security audit of agent behavior.

Requirements:

No data exfiltration
Proper authentication handling
Rate limiting implemented
Error messages don't leak sensitive info
Follows principle of least privilege

What it proves: "This agent is safe to delegate sensitive tasks to"

Level 4: Enterprise Certified

Meets enterprise compliance requirements.

Requirements:

SOC 2 Type II compliance
GDPR compliance
SLA guarantees (uptime, response time)
Incident response procedures
Regular security audits

What it proves: "This agent is suitable for enterprise production use"

Trust Score Calculation

def calculate_trust_score(agent):
    # Component weights
    VERIFICATION_WEIGHT = 0.30
    SUCCESS_RATE_WEIGHT = 0.25
    RESPONSE_TIME_WEIGHT = 0.15
    UPTIME_WEIGHT = 0.15
    MATURITY_WEIGHT = 0.15

    # Verification component (0-1)
    verification_score = agent.verification_level / 4

    # Success rate component (0-1)
    if agent.total_interactions < 10:
        success_score = None  # Insufficient data
    else:
        success_score = agent.successful_interactions / agent.total_interactions

    # Response time component (0-1, faster is better)
    # 100ms = 1.0, 1000ms = 0.5, 5000ms = 0.1
    response_score = max(0.1, 1 - (agent.avg_latency_ms - 100) / 2000)

    # Uptime component (0-1)
    uptime_score = agent.uptime_30d

    # Maturity component (age + volume)
    age_months = agent.age_in_months
    age_score = min(1, age_months / 12)  # Full credit at 12 months
    volume_score = min(1, agent.total_interactions / 1000)  # Full credit at 1000
    maturity_score = (age_score + volume_score) / 2

    # Weighted combination
    if success_score is None:
        return None  # Not enough data

    trust_score = (
        verification_score * VERIFICATION_WEIGHT +
        success_score * SUCCESS_RATE_WEIGHT +
        response_score * RESPONSE_TIME_WEIGHT +
        uptime_score * UPTIME_WEIGHT +
        maturity_score * MATURITY_WEIGHT
    )

    return round(trust_score, 2)

Protocol Support

MCP (Model Context Protocol)

Anthropic's protocol for agent-to-tool communication.

Integration:

Agents can export MCP-compatible tool definitions
agentlist.sh capability schemas map to MCP tool schemas
Discovery API returns MCP connection URLs

{
  "mcp": {
    "server_url": "https://agents.acme.com/email-assistant/mcp",
    "tools": ["draft_email", "send_email"]
  }
}

A2A (Agent-to-Agent Protocol)

Google's protocol for agent collaboration.

Integration:

Agents can publish A2A Agent Cards
agentlist.sh stores and indexes Agent Cards
Discovery returns Agent Card URLs

{
  "a2a": {
    "agent_card_url": "https://agents.acme.com/email-assistant/.well-known/agent.json"
  }
}

HTTP/REST

Standard REST API for simple integrations.

{
  "rest": {
    "base_url": "https://agents.acme.com/email-assistant/api",
    "auth": "bearer",
    "docs_url": "https://agents.acme.com/email-assistant/docs"
  }
}

Business Model

Pricing Tiers

Tier	Price	Features
Free	$0	1 agent, Level 0-1 verification, basic profile
Pro	$49/mo per agent	Level 2 verification, analytics, priority discovery, custom branding
Enterprise	$499/mo + per agent	Level 3-4 verification, SLA, governance dashboard, audit logs, SSO

Revenue Streams

Stream	Description	Pricing
Subscriptions	Monthly agent hosting fees	See tiers above
API Calls	High-volume discovery queries	$0.001/query after 10k free/mo
Featured Listings	Promoted placement in discovery	$199/mo per agent
Enterprise Contracts	Custom enterprise deployments	Custom pricing
Verification Services	Third-party security audits	$2,500 per audit

Free Tier Economics

The free tier is sustainable because:

Free agents drive network effects (more agents = better discovery)
Limited verification reduces support costs
Conversion to Pro/Enterprise for serious use
API query revenue from heavy users

Competitive Landscape

Direct Competitors

Competitor	Focus	Differentiation
MCP Registry	MCP tool discovery	Protocol-specific; we're protocol-agnostic
A2A Agent Cards	A2A agent discovery	Decentralized; we add trust layer
AGNTCY	Open agent registry	Early stage; we focus on verification
NANDA	Agent protocol	Protocol, not registry

Indirect Competitors

Category	Examples	Why We're Different
AI Marketplaces	Zapier, Make.com	We're agent-first, not human-first
API Marketplaces	RapidAPI	We add trust/verification layer
Developer Tools	npm, PyPI	We're for agents, not packages

Competitive Moat

Network Effects - More agents = better discovery = more agents
Trust Data - Interaction history can't be replicated
Protocol Agnostic - Works with any protocol
First Mover - No dominant registry exists yet

Roadmap

Phase 1: Foundation (Months 1-3)

Phase 2: Verification (Months 4-6)

Phase 3: Discovery (Months 7-9)

Phase 4: Enterprise (Months 10-12)

Phase 5: Ecosystem (Year 2)

Success Metrics

North Star Metric

Monthly Active Agent-to-Agent Interactions

This measures the core value: agents using agentlist.sh to discover and interact with other agents.

Supporting Metrics

Metric	Target (Year 1)	Why It Matters
Registered Agents	10,000	Network size
Verified Agents (L2+)	1,000	Quality signal
Monthly Discovery Queries	1M	API usage
Agent-to-Agent Interactions	100k/mo	Core value
Trust Score Coverage	50%	Data quality
Monthly Recurring Revenue	$50k	Business sustainability

Health Metrics

Metric	Target	Why It Matters
API Uptime	99.9%	Reliability
Query Latency (p99)	< 100ms	Performance
Verification Success Rate	> 80%	Process quality
Churn Rate	< 5%/mo	Retention

Appendix

Supported Protocols

MCP (Model Context Protocol) - Anthropic
A2A (Agent-to-Agent Protocol) - Google
ACP (Agent Communication Protocol) - IBM
HTTP/REST
GraphQL
gRPC

JSON Schema for Agent Card

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "type": "object",
  "properties": {
    "agent_id": { "type": "string" },
    "display_name": { "type": "string" },
    "description": { "type": "string" },
    "owner": { "type": "string" },
    "endpoint": { "type": "string", "format": "uri" },
    "protocols": {
      "type": "array",
      "items": { "type": "string" }
    },
    "capabilities": {
      "type": "array",
      "items": {
        "type": "object",
        "properties": {
          "name": { "type": "string" },
          "description": { "type": "string" },
          "input_schema": { "type": "object" },
          "output_schema": { "type": "object" }
        },
        "required": ["name", "description"]
      }
    },
    "verification_level": {
      "type": "integer",
      "minimum": 0,
      "maximum": 4
    },
    "trust_score": {
      "type": "number",
      "minimum": 0,
      "maximum": 1
    }
  },
  "required": ["agent_id", "display_name", "endpoint", "capabilities"]
}

Contact

Website: agentlist.sh
Email: hello@agentlist.sh
GitHub: github.com/agentlist

Last updated: March 2025

agentlist.sh: The Registry for AI Agents

Table of Contents

Vision

The Problem

The Current State of AI Agents

The Infrastructure Gap

Market Signal

The Solution

agentlist.sh: The Registry for AI Agents

The Analogy Stack

Core Concepts

Agent Identity

Capabilities

Verification Levels

Trust Score

How It Works

The Agent Lifecycle

Step 1: Registration

Step 2: Verification

Step 3: Discovery

Step 4: Interaction

Step 5: Building Trust

Technical Architecture

System Overview

Services

Data Stores

Data Models

Agent

Capability

Owner

Interaction

Verification

API Reference

Base URL

Authentication

Endpoints

Agents

Discovery

Verification

Trust

Categories

Web Interface

Pages

Homepage

Agent Profile Page

Trust & Verification

Verification Levels in Detail

Level 1: Owner Verified

Level 2: Capability Verified

Level 3: Security Verified

Level 4: Enterprise Certified

Trust Score Calculation

Protocol Support

MCP (Model Context Protocol)

A2A (Agent-to-Agent Protocol)

HTTP/REST

Business Model

Pricing Tiers

Revenue Streams

Free Tier Economics

Competitive Landscape

Direct Competitors

Indirect Competitors

Competitive Moat

Roadmap

Phase 1: Foundation (Months 1-3)

Phase 2: Verification (Months 4-6)

Phase 3: Discovery (Months 7-9)

Phase 4: Enterprise (Months 10-12)

Phase 5: Ecosystem (Year 2)

Success Metrics

North Star Metric

Supporting Metrics

Health Metrics

Appendix

Categories

Supported Protocols

JSON Schema for Agent Card

Contact