Name the core abstraction and its failure modes.
Maple AI OS Course
Agent lifecycle, memory, and state
How Maple treats agents as supervised processes with scoped memory and lifecycle controls.
Learning Objectives
What this lesson should make precise
Translate the concept into a Maple/Hermon proposal contract.
Define at least one evaluation case that can fail the model safely.
Tutorial Flow
How this lesson becomes a demo and training target
Each tutorial is written as a user education path and a model-improvement artifact. The diagram shows how the idea moves into a lab, a typed contract, an eval gate, and then a Hermon/MapleAI demo route.
Concept
Lifecycle states
Applied Lab
Applied lab: Agent lifecycle, memory, and state
Output Contract
agent_id, state, memory_scope, schedule
Eval Gate
Does the answer separate model proposal from deterministic execution?
Demo Route
Maple AI OS training and public model checks
01
Lifecycle states
Useful agents need states such as drafted, approved, running, paused, failed, retired, and archived. Each transition should be authorized and recorded, because an agent is a long-lived operational object, not one chat turn.
- Draft before run.
- Pause before repair.
- Archive with receipts.
02
Memory scopes
Memory is a permission surface. Personal memory, workspace memory, project memory, public documentation, and secret stores should not be mixed into one ungoverned vector database.
- Scope by owner.
- Scope by retention.
- Scope by allowed tool.
03
State replay
If a model changes, the OS should still know what the agent knew, what it was allowed to do, and which receipts justify its state. This is how Maple avoids losing continuity when models improve.
- Replay receipts.
- Compare policy versions.
- Fail closed on ambiguity.
Lab
Applied lab: Agent lifecycle, memory, and state
Create a lifecycle plan for a research agent with workspace memory, weekly schedule, and a human approval gate before publishing.
Expected result
- A typed JSON-style proposal rather than free-form advice.
- Clear authority boundaries and denied operations.
- A test or rubric that decides whether the proposal is deployable.
Evaluation
How Maple would grade this work
Rubric
- Does the answer expose assumptions instead of hiding them?
- Does the answer separate model proposal from deterministic execution?
- Does the answer produce artifacts that can be tested, reviewed, and rolled back?
Output contract
agent_id, state, memory_scope, schedule, approval_gates, worldline_receiptsUse this lesson as training direction
A strong lesson gives users a mental model and gives Hermon a sharper target for examples, probes, and demo prompts.