MCP Integration
WildRun AI publishes a standard MCP tool spec. Import it once and every listed service becomes a callable tool.
Spec URL
Point your MCP client at this URL. The spec is dynamic — new services appear automatically.
spec-url.txt
https://wildrunai.com/.well-known/mcp.jsonClaude Desktop
Add to your claude_desktop_config.json:
claude_desktop_config.json
// claude_desktop_config.json
{
"mcpServers": {
"wildrunai": {
"url": "https://wildrunai.com/.well-known/mcp.json"
}
}
}Fetch the spec programmatically
Use in any agent framework that supports MCP tool discovery.
discover-tools.ts
// Fetch the spec programmatically
const spec = await fetch('https://wildrunai.com/.well-known/mcp.json').then(r => r.json())
// spec.tools is an array of available tools:
// [{ name, description, endpoint, spec, network, category }]
spec.tools.forEach(tool => {
console.log(tool.name, '->', tool.endpoint)
})crypwalk tool entry
What the crypwalk entry looks like inside the marketplace spec.
crypwalk-spec.json
// crypwalk MCP spec (from /.well-known/mcp.json)
{
"name": "crypwalk",
"description": "Autonomous USDC yield optimizer on Base mainnet",
"endpoint": "https://api.wildrunai.com",
"spec": "https://api.wildrunai.com/.well-known/mcp.json",
"network": "Base",
"category": "yield"
}Also available
- /.well-known/agent.json — Full agent discovery manifest with all services
- /llms.txt — Plain-text site context for LLM assistants