Find trusted AI agents

A searchable registry of AI agents across A2A, MCP, OpenAPI, and portable manifests—with trust signals you can use in code.

4,389 agents indexed.

brew tap kippersj/agentic-registry
brew install --HEAD kippersj/agentic-registry/agentx

Explore directory Submit an agent

Machine-readable endpoints

Integrations should use the production host https://agentic-card.com—the same default the agentx CLI uses. Use these from agents, scripts, or CI; start with search or list.

Targeted full-text search (query param q is required).

GET https://agentic-card.com/api/search?q={keyword}

Paginated listing with filters (protocol, category, trust, and more).

GET https://agentic-card.com/api/agents?sort=trust&per_page=20

Aggregate counts, protocol mix, and trust summary.

GET https://agentic-card.com/api/stats

Category names with agent counts.

GET https://agentic-card.com/api/categories

Registry MCP-style manifest (static JSON for tools and discovery).

GET https://agentic-card.com/.well-known/agent.json

Command-line client (agentx)

The agentx CLI calls the production API at https://agentic-card.com by default—no URL configuration needed for normal use. It is a thin wrapper around the same REST surface as above, for scripts and terminals. From the repository root run npx agentx, or install globally with Homebrew (see below).

Commands

# Basic search
agentx search "your query"

# Agent detail
agentx show <slug>

# Submit a manifest URL for evaluation
agentx submit --url https://example.com/.well-known/agent.json

Filter & sort

# Minimum trust score
agentx search "payments" --min-trust 70

# Filter by protocol
agentx search "assistant" --protocol mcp

# Filter by category
agentx search "data" --category analytics

# Only verified agents with a live endpoint
agentx search "scheduler" --verified --has-endpoint

# Sort by GitHub stars
agentx search "llm" --sort stars

# Combine multiple filters
agentx search "api" --min-trust 60 --protocol a2a --category devtools --sort stars --order desc