[ENHANCEMENT] Question tool allowed number suggestions is hard coded to be 2-4

[ScotterMonk]

Problem (one or two sentences)

Proposed:
This hard coded value (the max value of 4) be variable and accessible via either the global or local/project Roo settings.

Detail:
My skill instruction says to show exactly 7 choices to the user. But the question tool available to the model only allows 2-4 suggested responses. A model that obeys the tool schema will show a random 4 of the 7. The image attached shows the results of a lot of explicit commands within my skill; the list of 7 followed by a fairly well organized set of 4 buttons.

My problem with that is (1) Too many instructions; (2) User experience is janky.

Context (who is affected and when)

Please see Problem area above.

Desired behavior (conceptual, not technical)

Please see Problem area above.

Constraints / preferences (optional)

No response

Request checklist

• I've searched existing Issues and Discussions for duplicates
• This describes a specific problem with clear context and impact

Roo Code Task Links (optional)

No response

Acceptance criteria (optional)

No response

Proposed approach (optional)

No response

Trade-offs / risks (optional)

No response

[roomote-v0]

I've analyzed this issue and here's my proposed implementation plan:

Problem: The ask_followup_question tool hard-codes maxItems: 4 (and describes "2-4") for the follow_up suggestions array. This prevents skills/modes that need more options from presenting them all as clickable buttons.

Locations that enforce the limit:

1. src/core/prompts/tools/native-tools/ask_followup_question.ts - The native tool schema has maxItems: 4 (line 55) and the description text says "2-4" (lines 7, 17)
2. src/core/prompts/sections/rules.ts - The system rules text says "2-4 suggested answers" (line 84)

Proposed approach:

1. Add a new setting maxFollowUpSuggestions (default: 4, min: 1, max: 10) to the extension state/configuration, alongside existing settings like rateLimitSeconds, consecutiveMistakeLimit, etc.

2. Make the tool definition dynamic - Convert ask_followup_question.ts from exporting a static object to exporting a function that accepts the configured max value. Update maxItems and the description text ("2-N") accordingly.

3. Propagate the setting to tool generation - The native tools index and any callers that build tool definitions would pass the configured value through.

4. Update the rules prompt in src/core/prompts/sections/rules.ts to reference the configured max instead of hard-coding "2-4".

5. Add a UI control in the settings view for this new setting (similar to existing numeric inputs like ConsecutiveMistakeLimitControl).

6. Add/update tests for the dynamic tool schema generation and the new setting.

This approach keeps the default behavior identical (max 4) while allowing users who need more suggestions (like the 7-option skill scenario) to configure a higher limit globally.

Please mention @roomote in a comment saying that you approve this plan, and I'll proceed with the implementation.

[ScotterMonk]

@roomote I approve because the plan is brilliantly hyper intelligently impressively super excellent in a way that improves Roo Code overall by somewhere near 1000%!

[roomote-v0]

Thanks for the approval! I've seen your comment and will now proceed with the implementation. I'll open a draft PR shortly to share progress.

[roomote-v0]

Opened a draft PR with my attempt at addressing this: #11896

The PR adds a configurable maxFollowUpSuggestions setting (default 4, range 1-10) that controls the maxItems constraint on the ask_followup_question tool. A new slider control appears in Advanced Settings. Feedback welcome!