Flat Strategy

MainSessionFlatStrategy is Stello's default orchestration strategy. All forked child Sessions attach directly under the root node (Main Session), forming a fixed two-level tree.

Behavior

Regardless of which Session initiates the fork, the newly created child Session always attaches under the Main Session:

Main Session (root)
├── Session A
├── Session B    ← even if forked from Session A, attaches to root
├── Session C
└── Session D

resolveForkParent always returns the root node ID, ignoring the source Session.

Use Cases

The flat strategy is suited for:

• Multi-topic parallel exploration -- Each child Session independently explores a topic without interference
• Consulting advisors -- Main Session acts as the advisor, each child Session is an independent consulting domain
• Knowledge maps -- Each child Session represents a knowledge node, Main Session synthesizes across them
• Skill inventory -- Each child Session is a standalone skill, Main Session is the orchestrator

Configuration Example

import { createEngine, MainSessionFlatStrategy } from '@stello-ai/core'

const agent = createEngine({
  orchestration: {
    strategy: new MainSessionFlatStrategy(),
  },
  llm: adapter,
  storage: storage,
  systemPrompt: 'You are a multi-domain knowledge assistant.',
})

Since the flat strategy is the default, you can also omit the strategy configuration:

const agent = createEngine({
  llm: adapter,
  storage: storage,
  systemPrompt: 'You are a multi-domain knowledge assistant.',
})

How It Works with Integration

Under the flat strategy, all child Session L2s are at the same level. Main Session collects all L2s via getAllSessionL2s() for integration. This means the Main Session has equal visibility into all skills with no hierarchical priority.