Most teams ship agent demos; few ship agents that survive production. The AgenticProduct ecosystem is our answer to that gap: the Agentic Product Standard defines the contract — six principles, the autonomy ladder, the harness layers and eval discipline — and six open-source products implement its surfaces so you don't have to build them from scratch. AgenticOps runs the fleet; AgenticMind gives agents auditable knowledge & memory; AgenticPerformance measures every run with traces and evals; AgenticSelfHealingCode repairs what breaks; AgenticGateway is the model & cost plane every LLM call passes through; and AgenticAssurance red-teams any agent in the loop. Each product runs standalone and integrates with its siblings through optional adapters — never hard dependencies — and the whole family comes up locally with one command.
Seven members, one loop: run → remember → measure → heal — with the model plane and the assurance plane across all of it.
The canonical open standard: six principles, the autonomy ladder, composition patterns, the harness layers, eval discipline and a self-assessment scorecard every member of the family conforms to.
The runtime & operations layer: agents as versioned manifests, coordinated scheduling without double-runs, a durable backlog, bounded execution and fleet observability.
The knowledge & memory layer: auditable, citation-enforced answers with a replayable why-trace and a self-improving corpus, served over MCP.
The evals & observability layer: OTel traces, per-agent golden-set evals with a CI gate, named failure clusters and a governed improvement loop.
The self-healing ops layer: grounded incident diagnosis, signed signal intake and repair actions on earned autonomy, with a kill switch.
The model & cost plane: one OpenAI-compatible key for the fleet, eval-measured routing, fail-closed cost ceilings and an evidence trail.
The security layer: red-teams any agent against the OWASP Agentic Top 10 and MITRE ATLAS, finds toxic tool-flow paths and gates CI with SARIF output.
To help you with any questions that are not listed here, we offer free, no-obligation consultations
No. Every product runs standalone and earns its place on its own; integrations between siblings are optional adapters, never hard dependencies. Start with the surface that hurts most — operations, memory, measurement or security.
Yes — every repository is public under Moai-Team-LLC on GitHub, licensed Apache 2.0 or MIT. Free to use, self-host and build on, including commercially.
AgenticOps runs the fleet, AgenticMind remembers, AgenticPerformance measures, AgenticSelfHealingCode heals. AgenticGateway is the model plane every LLM call in that loop passes through, and AgenticAssurance red-teams the whole thing — all of it conforming to the Standard.
Read the Standard and score your product against its scorecard — or bring the whole family up locally with the one-command family script from the standard repository, then adopt the layers that close your biggest gaps.
That's exactly what our agentic software development service is: we design and ship production-grade agentic products on this stack — architecture, harness, tools, memory and evals — and you keep an open-source foundation with no lock-in.
More from our agentic practice — the standard we build to, the open-source layers that implement it, and the team that ships it.
Want a production-grade agent stack without betting on a black box? Tell us where you are and we'll map the ecosystem onto your product.