The default directory of
AI agent functions
Searchable, trust-scored directory of MCP-compatible functions. Install in one click for Claude Desktop, Cursor, and VS Code.
Everything you need to build with agents
The Function Registry gives AI agents access to verified, trust-scored functions they can call with confidence.
Trust Scores
Every function has a trust score computed from real execution outcomes. Functions that run successfully earn higher scores.
Sandbox Execution
Every function runs in an isolated WebAssembly sandbox. No filesystem access, no network egress unless explicitly granted.
Verification Levels
From automated L1 checks to platform-signed attestation at L4, each function's trust tier is clearly displayed.
One-Click Install
Add functions to your MCP server configuration with a single click. Works with Claude Desktop, Cursor, and VS Code.
Verification levels explained
Automated
Automated malware scan and schema validation. Every function starts here.
Malware + DRE + FXCERT
Security scan, data-risk evaluation, and platform-signed execution certificates (FXCERT).
Full + Manual Review
All L2 checks plus manual review of safety-relevant aspects.
Certified
Platform-verified with Ed25519-signed attestation. Highest trust.
Browse by category
Get started in minutes
Install the MCP server
Add FunctionFly to your MCP server configuration:
{
"mcpServers": {
"functionfly": {
"command": "npx",
"args": ["-y", "@functionflycom/mcp-server"],
"env": {
"FUNCTIONFLY_API_KEY": "ffp_your_api_key_here"
}
}
}
}Browse the registry
Explore trust-scored functions and add them to your agent configuration with one click.
Start building
Your agent can now call verified functions with full trust scores and execution tracking.
Ready to build with
trust infrastructure?
Join thousands of developers building reliable AI agent applications.