// Developers
Open-source primitives for agentic systems.
The developer surface of AgentLayer Platform. Reference implementations, four @agentlayer/* pattern packages, SDKs, and certifications. Engineering teams can engage these primitives without ever talking to us.
// Pattern Packages
Four primitives, extracted from production code.
Patterns proven against real regulated-industry workloads, then absorbed into the engine that powers every AgentLayer deployment. Open-sourced where they can be. Halal policy enforcement and Malaysia-regulatory audit trail stay closed.
@agentlayer/orchestrationIn preparationAgent loop, tool dispatch, multi-step planning, lifecycle and state handoff. The orchestration substrate.
@agentlayer/identityIn preparationPrincipal-vs-agent action delegation, KYA (Know-Your-Agent), credential scoping.
@agentlayer/auditIn preparationAppend-only action logs, attribution, replay, citations on retrieval. Regulator-aware audit trail.
@agentlayer/observabilityIn preparationHeartbeat health, task lineage, session state, agent-action traceability.
// Reference Implementations
Production reference. Live operator surface.
@agentlayer/agent-kevin is a production agentic system running daily, validating the engine's patterns before they deploy at enterprise scale. @agentlayer/studio is the platform's operator surface, live as an interactive prototype today at studio.agentlayer.one.
▸ @agentlayer/agent-kevin
Personal AI assistant — canonical internal-team-agents reference: knowledge management, task orchestration, daily briefings, autonomous heartbeat tasks, plugin dispatch. Open source and running in production daily, available at github.com/AgentLayer1/agent-kevin.
▸ @agentlayer/studio
AgentLayer Studio — the operator surface of the platform. Activity feed, agents, kanban + list task views, timeline, audit trail, approvals, notifications, light/dark theming, drag-and-drop, real-time streaming. Live interactive prototype at studio.agentlayer.one. Closed commercial component; engine integration shipping in Year 1.
// Engineering Principles
▸ Orchestration over framework
Patterns that compose with the language and tooling teams already use. No framework lock-in.
▸ Audit trail by default
Every agent action attributable, replayable, reviewable.
▸ Observability first
Heartbeat health, task lineage, session state. All surfaced.
▸ Human-in-the-loop is a feature
Designed-in approval points, not bolted-on guardrails. A core capability, not a fallback.