AgentPlat

AgentPlat is an open-source TypeScript framework for persistent human-agent collaboration, with shared artifacts, human approvals and controlled execution on your infrastructure.

Apache-2.0. AgentPlat 1.1.0 is published under npm latest. Your application supplies UI, verified identity, policies, model providers and infrastructure. Review component maturity and release channels. Technical architecture designed by Grishen (opens in a new tab).

New in 1.1.0

Understand purpose, inceptions, signals, references, limits and budgets with one support example.

Keep existing instruction-driven agents or explicitly enable governed purpose execution. Owners configure limits, budgets and activation; mission outcomes require evidence. Both modes can coexist. Start with the 1.1 adoption guide, support demonstration and verified release record. Jev remains an optional assessor.

Start here

• When to use AgentPlat and alternatives

• Build persistent human-agent collaboration in TypeScript

• Add human approval to a multi-agent workflow

• Resume coordination after a process failure

• Inspect reproducible evidence

• Install AgentPlat and run your first agent

• Understand the core concepts

• Read the architecture

• Explore Agent Rooms

• Evolve a collective with Agent Morphogenesis

• Run durable Room coordination

• Follow the operational Agent Rooms quickstart

• Review the Agent Rooms Beta 5 release

• Read the Agent Rooms paper (opens in a new tab)

• Run your first collective

• Learn how recovery works

• Integrate Agent Mesh

• For AI coding agents

• Connect an MCP client to your own runtime

What AgentPlat provides

AgentPlat combines agent execution with the operational controls required for dependable multi-agent systems:

• Bounded coordination: peers coordinate through local, policy-approved views rather than an unbounded global broadcast.
• Recoverable execution: assignments, teams and missions can recover from peer, link and capability failures.
• Governed organizational evolution: Agent Morphogenesis can change agents, roles, Teams, strategies and organizational epochs without bypassing existing authority owners.
• Controlled authority: planning, coordination and external effects are separate authority boundaries.
• Verifiable operations: decisions, recoveries, protected actions and outcomes produce inspectable evidence.

AgentPlat does not make autonomy magically safe or guarantee a specific scale in every environment. Behavior depends on topology, policy, model adapters, persistence and deployment conditions. The documentation separates implemented capability, validation evidence and operational limitations.

Explore the platform

• Collective Runtime
• Agent Rooms
• Agent Mesh
• Agent Morphogenesis
• Safety and Governance
• Evidence and Operations
• Scale and Evaluation
• Package and API Reference
• MCP Runtime (self-hosted)

For machine-readable context, see /llms.txt and /ai/context.md.