Version 2.0.69

What: Fixed the prompt suggestion feature's maxOutputTokens limit, which was set far too low at 50 tokens, potentially causing truncated or failed suggestions.

Details:

• The maxOutputTokens was increased from 50 to 16,000 tokens
• This provides the model adequate headroom to generate suggestions without hitting token limits
• The prompt still requests brief 3-8 word suggestions, but the model now has sufficient capacity to reason and respond reliably
• Evidence: nH8() function at line 366252 - maxOutputTokens changed from iH8 (value: 50) to hardcoded 16000

What: The prompt suggestion function was refactored for better code organization and maintainability.

Details:

• The system prompt text was extracted from inline to a module-level constant (iH8 at line 366298)
• The function was converted from promise-chain style to explicit async/await
• Response parsing changed from .then() chain with .find() to a for...of loop with early return
• Variables were extracted for better readability (cacheSafeParams to Q, canUseTool callback to B)
• Evidence: nH8() at lines 366252-366273 in v2.0.69 vs lines 366252-366280 in v2.0.68

What: Several import statements were reorganized and consolidated across the codebase.

Details:

• Node.js built-in imports (node:fs, node:child_process, node:path, node:os, node:util, node:process) were restructured
• Crypto imports consolidated to use named imports (randomBytes, randomUUID, timingSafeEqual)
• Stream import changed from default to named import (PassThrough)
• These are purely organizational changes with no functional impact
• Evidence: Import changes at multiple locations throughout the codebase (see diff lines 10-72, 76-170)