Open-source AgentOps control plane for enterprise AI agents.
Govern, evaluate, route, approve, audit, and benchmark enterprise AI agents before production. Built as a deterministic control-plane reference implementation with live connectors and live providers disabled by design.
A control plane for the agent lifecycle.
Intake
Capture business intent, owner, domain, outcome, data, and autonomy expectation.
Govern
Classify suitability, risk, data readiness, controls, approvals, and production gates.
Enforce
Apply policy-as-code to tools, data, environments, model routing, identity, and secrets.
Observe
Record traces, audit events, decisions, blocked actions, evidence, and readiness reports.
Launch
Use benchmark, release-evidence, deployment, public-site, and launch-candidate checks.
Explore the public launch candidate.
Interactive Demo Path
Follow the procurement-agent control-plane journey from intent to readiness report.
Control Plane Console
Review the unified capability map, API surface, release status, and demo flow.
API Catalog
Inspect deterministic endpoint groups and OpenAPI-lite metadata.
Benchmark Console
View reusable scenarios, benchmark suites, scoring posture, and sample runs.
Release Evidence
Review validation snapshot, public proof bundle, and demo recording readiness.
Launch Candidate
Check GitHub Pages finalization, publication sequence, evidence, and social launch copy.
Procurement Agent Accelerator
Demonstrates an end-to-end control-plane flow: PO, invoice, challan, vendor consistency, governance, policy, runtime, sandbox tool execution, traceability, and readiness reporting.
Live execution remains off.
The launch candidate does not call live providers, execute live connectors, store raw secrets, provision cloud infrastructure, or implement production IAM. It models the control plane before production.
v2.8 is the closure point for public launch.
Further feature development should pause until the repository is published, the demo is recorded, and external feedback is collected.