[FEATURE] "Claude too slow?" button — sub-agent that analyses context and recommends speed settings

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

Response speed is a pain point for me as a heavy daily user (Max 5x/20x, multiple hours every day). Some of this is likely my own suboptimal tuning of effort level, the thinking toggle, and context hygiene — but most users won't read the documentation or have the inclination to learn how to tune these settings. The result is people running high effort + thinking on mechanical tasks that don't need it, and waiting needlessly.

Proposed Solution

A button (Claude Code VS Code extension and claude.ai) labelled "Claude too slow?" that triggers a sub-agent to analyse the current chat setup and context, then recommend faster settings.

Example output:
"For this complexity of work you'll do better with medium effort and thinking off — the task is mechanical, so this should speed up responses. Want me to set it for you? Y/N"

The user confirms and the settings apply. This puts effort/thinking/model tuning within reach of users who would never configure it manually.

Alternative Solutions

Current workarounds:

• Manually adjusting effort level and the thinking toggle — requires knowing they exist and what they do; most users don't.
• Reading the documentation — most users won't.
• Switching models manually — trial and error, no guidance on what fits the task.

None of these guide the user; they assume prior knowledge and initiative. An assistant that reads the context and recommends settings removes that barrier.

Priority

Medium - Would be very helpful

Feature Category

CLI commands and flags

Use Case Example

1. I'm doing mechanical, low-complexity work (renaming variables, simple edits) but have high effort + thinking on from an earlier complex task.
2. Responses are slow and I don't notice the settings mismatch.
3. I click "Claude too slow?".
4. A sub-agent inspects the current task and context and replies: "This task is mechanical — medium effort, thinking off will be faster. Set it? Y/N".
5. I confirm; responses speed up immediately without my needing to know which knobs to turn.

Additional Context

• Environment: Claude Code VS Code extension + claude.ai (browser); Opus 4.8; Max 5x / 20x subscription.
• Even shipped as an opt-in beta, features like this would make Claude meaningfully more user-friendly for the majority who never touch advanced settings.