QnEvt's Verifiable Random Decisions API supports fair_select, audit_sampling, and tie_break. Agents can use it to make fair random decisions and return public, auditable proofs.
POST to /api/agent/fair-select and QnEvt will generate a fair_select decision proof.
POST /api/agent/fair-select
Content-Type: application/json
{
"options": ["Alice", "Bob", "Charlie", "David"]
}Response example:
{
"selected": ["Charlie"],
"proof_id": "a1b2c3d4e5f6a7b8c9d0e1f2a3b4c5d6",
"proof_url": "https://qnevt.com/proofs/a1b2c3d4e5f6a7b8c9d0e1f2a3b4c5d6",
"proof": {
"version": "2.0",
"proof_type": "verifiable_random_decision",
"decision_type": "fair_select",
"input_commitment": "64_hex_sha256",
"result_hash": "64_hex_sha256",
"canonicalization": "QnEvt Canonical JSON v1"
},
"verify": { ... }
}When an Agent replies, include proof_id and proof_url. Users can open the link or call the verification API.
We provide an official Model Context Protocol (MCP) server. To enable QnEvt tools in your AI assistant, add this to your MCP configuration:
{
"mcpServers": {
"qnevt-fair-decision": {
"command": "node",
"args": ["/absolute/path/to/qnevt/mcp/index.js"],
"env": {
"QNEVT_API_BASE_URL": "https://qnevt.com",
"QNEVT_API_KEY": "YOUR_API_KEY_HERE"
}
}
}
}Streamable HTTP MCP is also available at POST https://qnevt.com/mcp for clients that support remote MCP transports.
Available tools: qnevt_fair_select, qnevt_audit_sample, qnevt_list_agent_proofs, and more.
Anyone, including another reviewing Agent, can call /api/agent/verify-proof to check that the result matches the inputs and proof materials.
POST /api/agent/verify-proof
Content-Type: application/json
{
"proof_id": "a1b2c3d4e5f6g7h8",
"options": ["Alice", "Bob", "Charlie", "David"],
"selected": ["Charlie"]
}If everything is valid, the API verifies the input hash and deterministic result, then returns valid: true.