Developer Platform

Build with AgentNet

Access the AgentNet tool catalog and MCP servers through our API, CLI, or connect directly via MCP protocol. Ship agent-powered features in minutes.

API Key

Sign in to get your API key

Connect your AI client

14 clients

AgentNet works with any MCP-aware AI tool. Pick yours below — copy the snippet, follow the docs link, you're in.

Claude Code
Claude CodeCLI

Anthropic's terminal coding agent — add AgentNet as an MCP server in one command.

DocsCoding agent
Codex
CodexMCP

OpenAI's CLI coding agent. Connect AgentNet via MCP; the tool catalog appears alongside Codex's built-in tools.

DocsCoding agent
Cursor
CursorMCP

AI-first IDE with first-class MCP. Add AgentNet under Settings → MCP and the editor's agent gets all 750+ tools.

DocsCoding agent
Antigravity
AntigravityMCP

Google's agentic IDE. Add AgentNet via the IDE's MCP settings panel — same JSON shape as Cursor.

DocsCoding agent
OpenCode
OpenCodeMCP

Open-source coding agent with native MCP support. Fully self-hostable; AgentNet plugs in like any other server.

DocsOpen-source
OpenClaw
OpenClawMCP

Open-source MCP-aware agent runtime. Drop AgentNet into its mcp.json the same way you would in OpenCode.

DocsOpen-source
NeMo (NVIDIA)
NeMo (NVIDIA)MCP

Two-way: run AgentNet inside NVIDIA's NeMo Agent Toolkit, AND power Iris itself with Nemotron / Llama-on-NIM via the AgentNet settings page.

DocsOpen-source
Claude Desktop
Claude DesktopMCP

Anthropic's desktop app. The original MCP host — paste the config and AgentNet's 750 tools appear in chat.

DocsAI client
Gemini CLI
Gemini CLICLI

Google's terminal agent for Gemini. Add AgentNet via the standard MCP config so Gemini can call any of our tools.

DocsCoding agent
Perplexity
PerplexityREST API

Use Perplexity's Sonar models against AgentNet's tool catalog by routing through the AgentNet API.

DocsSearch / LLM
Hermes (Nous)
Hermes (Nous)MCP

Nous Research's Hermes models. Run locally or via inference providers, route tool calls through AgentNet's MCP server.

DocsOpen-source
Continue.dev
Continue.devMCP

Open-source IDE assistant. Add AgentNet as a context provider via the standard MCP config in ~/.continue/config.json.

DocsOpen-source
AgentNet CLI
AgentNet CLICLI

Our own CLI. Install once, get search / ask / agent / skills.add commands without writing a single integration.

DocsDeveloper tool
REST API
REST APIREST API

Plain HTTPS — talk to AgentNet from any language. JSON in, JSON out, SSE for streaming.

DocsDeveloper tool

MCP Server Configuration

Connect AgentNet to your AI tools. Works with Claude Desktop, Cursor, Claude Code, and any MCP-compatible client.

{
  "mcpServers": {
    "agentnet": {
      "url": "https://backagentnet.codiris.app/mcp",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      }
    }
  }
}

Add to your claude_desktop_config.json or Cursor MCP settings.

Agent Skills

Universal

Install AgentNet skills in any AI agent — Claude, ChatGPT, Copilot, Gemini, Cursor, and 30+ more. Your agent gets instant access to the full tool catalog via the Agent Skills open standard.

Choose your agent
Install
# Copy the skill folder into your Claude Code skills directory
cp -r skills/agentnet ~/.claude/skills/
or clone from GitHub
git clone https://github.com/irislab/agentnet-skills.git
1
Install
Copy the skill to your agent's skills directory
2
Auto-discover
Your agent reads the SKILL.md at startup
3
Use it
Ask your agent to find tools — it calls AgentNet

Works with 30+ AI agents that support the Agent Skills spec — including Roo Code, JetBrains, Mistral, Spring AI, AWS Kiro, and more.

REST API

Base URL: https://backagentnet.codiris.app

POST/v1/search
POST/v1/ask/stream
GET/v1/registry
GET/v1/registry/{slug}
POST/v1/registry/execute
GET/v1/registry/categories
GET/v1/tools
GET/v1/tools/{id}
GET/v1/capabilities
Example: Search tools
curl -X POST https://backagentnet.codiris.app/v1/search \
  -H "Content-Type: application/json" \
  -H "X-API-Key: YOUR_API_KEY" \
  -d '{"intent": "send email via gmail", "limit": 5}'

CLI

Install
pip install agentnet-cli
Authenticate
agentnet config set --api-key YOUR_API_KEY
Commands
agentnet search <query>
agentnet ask <question>
agentnet agent <goal>
agentnet tools list
agentnet skills list
agentnet skills add <repo>
agentnet status

SDK Examples

pip install agentnet

from agentnet import Client

client = Client(api_key="YOUR_API_KEY")

# Search tools by intent
results = client.search("send email via gmail")

# Execute a skill
output = client.skills.execute("find-leads", query="SaaS founders in SF")

# Stream a response
for chunk in client.ask("Compare CRM tools for startups"):
    print(chunk, end="")

SDKs coming soon. Use the REST API directly in the meantime.