v1.4.487 v1.4.488
low risk
danielmiessler/FabricSep 29, 2026 → Sep 29, 20260 days between releases
What's New
Adds --maxTokens flag and fixes Claude 5 thinking support.
New Features
- Added --maxTokens CLI flag to cap model output tokens.
- Added shell completion support for the --maxTokens flag.
Improvements
- Fixed Anthropic thinking support on Claude 5 models (Sonnet, Opus, Fable).
- Preserved numeric thinking budgets on adaptive models for consistency.
Bug Fixes
- Fixed HTTP 400 errors on Claude 5 models when using the --thinking flag.
Upgrade Notes
The new --maxTokens flag is opt-in; setting it to 0 preserves existing defaults. Existing behavior remains unchanged unless the flag is explicitly used. Claude 5 models now correctly handle thinking budgets.
The Two Releases
v1.4.487
v1.4.487
Sep 29, 2026· 23 days ago
Adds convenient CLI flags to extract specific content from LLM responses, specifically code blocks.
Full release notesv1.4.488
v1.4.488
Sep 29, 2026· 23 days ago
Enhances CLI control for Anthropic models by adding token limits and fixing thinking support for Claude 5.
Full release notes