AgentPlatThe AgentPlat library
Practical guides to persistent collaboration, human governance and reliable agent execution.
Start here · Agent Rooms
Learn how persistent human-agent collaboration connects tasks, versioned artifacts and human decisions in an AgentPlat Agent Room.
Read the guideChoose a work model and plan your integration.
Foundations
Choose whether AgentPlat fits your TypeScript application by examining shared work, review requirements, deployment ownership and simpler execution paths.
Read articleFoundations
Understand the responsibilities of Agent Rooms, Collective Runtime, Agent Mesh, Morphogenesis and the adapters that connect AgentPlat to your infrastructure.
Read articleFoundations
Plan an Agent Rooms deployment around verified identity, persistent storage, provider adapters and operational checks, starting from the self-hosted reference app.
Read articleFoundations
Compare Room collaboration, coordination workers and durable workflow engines, including the role of the optional AgentPlat Rooms Temporal adapter.
Read articleConnect people, tasks and versioned work in a shared Room.
Agent Rooms
Learn how persistent human-agent collaboration connects tasks, versioned artifacts and human decisions in an AgentPlat Agent Room.
Read articleAgent Rooms
Understand Agent Room lifecycle transitions, read-only archives, subroom completion constraints and the records that explain progress across a shared objective.
Read articleAgent Rooms
Learn how typed AgentPlat Handoffs transfer context and ownership, and how progressive plans materialize eligible work without duplicating logical steps.
Read articleAgent Rooms
Understand durable human contribution requests, scoped knowledge bundles and external work-management projections while keeping Agent Rooms authoritative.
Read articleAgent Rooms
Understand revisioned execution sessions, durable intervention requests and provider checkpoints for controlling an active Agent Room run.
Read articleAgent Rooms
Learn how to integrate the Agent Rooms HTTP API with injected services, authenticated tenant context, response envelopes and revision-aware client behavior.
Read articleAgent Rooms
Understand Agent Room LiveView, composite cursors and server-sent events so your interface can reconnect without treating a stream as authoritative state.
Read articleAgent Rooms
Plan coordinated Agent Rooms schema upgrades, including migration order, operational projections, worker rollout and restore checks for PostgreSQL deployments.
Read articleAgent Rooms
Understand stable agent identities, published revisions and runtime-profile binding so an Agent Room run can be traced to the exact agent definition used.
Read articleCoordinate independent peers with explicit evidence and membership.
Agent Mesh
Learn when Agent Mesh is useful, how it coordinates independent runtimes and which transport, membership and effect-control responsibilities remain with your host.
Read articleAgent Mesh
Understand bounded Mesh envelopes, sender and scope validation, membership epochs and compatibility checks before peer messages influence agent coordination.
Read articleAgent Mesh
Explore causal anti-entropy, predecessor evidence and readiness gates in Agent Mesh, and why synchronized state does not authorize an agent action.
Read articleAgent Mesh
Learn how Agent Mesh uses governed membership epochs for enrollment, replacement and key rotation, and why old credentials do not establish current authority.
Read articleAgent Mesh
Understand bounded active and reserve peer views, policy-governed adaptation and how to evaluate sparse Agent Mesh coordination without assuming scale guarantees.
Read articleAgent Mesh
Understand how Agent Mesh and Collective Runtime handle incomplete peer views, partition policies, reconciliation and restrictions on irreversible effects.
Read articleAgent Mesh
Learn how the AgentPlat Rooms Mesh bridge projects objectives and work, returns results and preserves the separation between Room policy and Mesh authority.
Read articleControl execution and understand recovery, approvals and audit records.
Human governance
Understand artifact review, task approval and execution checkpoints, and where to enforce human approval in an AI agent workflow.
Read articleReliable execution
Explore stable work identity, durable results and lease-based recovery for AI agent workflows, including the limits of retrying external effects.
Read articleGovernance & reliability
Understand the checks that belong before protected agent effects, including current assignment, policy, fencing, approvals and effect-specific idempotency.
Read articleGovernance & reliability
Learn where AgentPlat inference controls constrain model invocation, how they relate to provider checkpoints and why calls outside the integration path remain separate.
Read articleGovernance & reliability
Learn what to retain in an AgentPlat audit trail, how causal checkpoints support replay and why reconstructing decisions must not blindly repeat external effects.
Read articlePlan missions, preserve continuity and govern organizational change.
Collective systems
Understand how AgentPlat Collective Runtime connects mission intent, planning, teams, execution and recovery through explicit policy-bound lifecycle transitions.
Read articleCollective systems
Distinguish plans, eligible assignments and effect authority in AgentPlat, including how expiry and reassignment preserve the reason work changed owners.
Read articleCollective systems
Learn how AgentPlat continuity binds checkpoints to missions, assignments and membership, and what must be revalidated when a replacement peer resumes work.
Read articleCollective systems
Understand AgentPlat Morphogenesis, including organizational proposals, scoped decisions, execution authorization and receipts from the subsystems that own changes.
Read articleEvaluate the framework with observable tasks and honest evidence.
Adoption & evaluation
Plan an AgentPlat evaluation with recorded prerequisites, concrete collaboration tasks, honest failure reporting and separate evidence for usability and software behavior.
Read articlePublished by AgentPlat, the open-source TypeScript framework for persistent human-agent collaboration. These guides explain design decisions and documented examples, with links to primary sources. Illustrative scenarios are identified as examples; documented checks are not presented as production guarantees.
Implementation details live in the documentation and source repository. Each article includes publication and, when revised, update dates.