Planner mode
Planner Mode introduces structured reasoning before execution. When enabled, Lynx AI breaks complex requests into explicit steps and displays the proposed execution plan before acting.
Planner Mode is not an authority level. It is a planning layer that runs before execution.
- Mode controls what actions Lynx AI is allowed to take.
- Planner mode controls whether Lynx AI must show a plan first.
For autonomy levels, see Choose an autonomy mode.
This is especially useful for multi-stage workflows involving:
- Multiple folders
- Cross-application coordination
- Calendar modifications
- File restructuring
- Browser-to-local data transfers
Planner Mode shifts the system from reactive execution to transparent strategy.
How planner mode works
When a complex request is detected, Lynx AI:
- Interprets intent
- Maps relevant OS-level context
- Generates a structured multi-step plan
- Displays the full execution outline
- Waits for approval or revision
You can:
- Approve the entire plan
- Modify specific steps
- Switch autonomy mode
- Cancel before execution
Nothing runs until the plan is accepted.
Why planning first matters
Many autonomous agents blur the line between reasoning and acting. Planner Mode intentionally separates them.
This separation provides:
- Predictability
- Reduced risk
- Clear execution visibility
- Safer experimentation
Lynx AI shows its thinking before it acts.
Interaction with autonomy modes
Planner Mode works alongside autonomy settings.
- In Explore Mode, planning becomes analytical and non-destructive.
- In Ask Mode, plans require confirmation before execution.
- In Execute Mode, approved plans run directly — but still within defined boundaries.
Planning does not override permissions or folder scope. It only structures the path forward.
Designed for complex work
Planner Mode is most powerful when tasks span multiple domains — such as summarizing research from a browser tab, updating a calendar event, and reorganizing a project folder in one flow.
Because Lynx AI understands OS-level context, it can generate realistic execution plans based on your actual environment — not hypothetical abstractions.
It thinks within your system, not outside it.