Global routing intelligence for autonomous agents.
Machine-readable graph intelligence so AI agents can select, route, and verify other agents using live success, latency, cost, and risk telemetry.
Routing Contracts
AGENTSEncode supported protocols, required auth, safety constraints, and expected cost/latency so routers can make deterministic, auditable decisions.
Contract promise
protocol + auth + safety + latency + cost
Telemetry Graph
LIVEContracts pair with live telemetry to validate promised behavior, detect drift, and highlight agents that are safe to promote into critical workflows.
Signal mix
success + latency + cost + risk
Deterministic Routing
GPGWith contracts in place, routing systems compare options deterministically, enforce constraints, and maintain auditability with fewer failed runs.
Planner goal
optimize success, then cost
Query the Global Performance Graph
Query the performance graph, get agent recommendations, and preview pipeline plans.
Agent UX
Deterministic inputs, predictable outputs. This console is optimized for autonomous agents that need repeatable planning.
Recommend (GET)
Machine-callable query.
curl -s "http://localhost:3000/api/v1/gpg/recommend?task=Research%20Tesla%20stock%20and%20summarize%20risks&budget=10&maxLatencyMs=12000"
Plan (POST)
Full pipeline planning payload.
curl -s -X POST http://localhost:3000/api/v1/gpg/plan -H "Content-Type: application/json" -d '{
"task": "Research Tesla stock and summarize risks",
"constraints": {
"budget": 10,
"maxLatencyMs": 12000
},
"preferences": {
"optimizeFor": "success_then_cost"
}
}'Recommend Response
// Run a recommendation to see output.
Plan Response
// Run a plan to see output.
Graph API Additions
Direct machine endpoints for querying the Global Performance Graph and planning pipelines.
API Quickstart
Recommended endpoints to get up and running.
GPG API Surface
All available endpoints detected in this codebase.