VT Code

What's Changed

0.84.0 - 2026-02-28

Highlights

Bug Fixes

• Strip x-extension fields from tool schemas and handle empty safety blocks (7ff0b66f)

Features

• Add terminal notifications for agent events and syntax highlighting for bash commands in TUI (d54ba169)

Other Changes

Other

• Update badge links in README.md (b74377e3) (@vinhnx)
• Merge pull request #607 from LiGoldragon/mentci-gemini-json-fix (db195063) (@vinhnx)
• Refactor code for improved readability and performance; simplify conditionals and remove unnecessary clones (93989d58)
• Add notification system with configurable preferences and delivery modes (e3be6b6b)
• Remove deprecated tool aliases and clean up related code for user input handling (5a23ab29)
• Unify ModelId and Provider across vtcode-config and vtcode-core; add regression tests to prevent future drift (3bef6a4d)
• Remove deprecated model ID and provider configurations, streamline model handling and update references across the codebase. (ebd573ec)
• Improve (d8ec16e4)
• Refactor code for improved readability and maintainability; streamline model handling and enhance diff preview functionality. (58607c7b)
• Add Qwen3.5-397B-A17B model support and update related identifiers and descriptions (a9e62f9c)
• Add turn elapsed timer feature and configuration options (f6867385)
• Remove xlsx skill and related files; update turn outcome message format (c31af51e)
• Implement exact token counting for various LLM providers and enable token counting feature (136c1b1f)
• Remove unused question-related policies and enhance token usage logic in ContextManager (4e39398a)
• Remove xAI Grok models and related configurations from the codebase (7b069620)
• Enhance file reference extraction to handle npm scoped packages and improve path validation (00646152)
• Enhance UI styling for PTY output and improve color handling across components (36cc1ca0)
• Add sidebar, slash command palette, and transcript widgets; implement session options and host adapter traits (fdc70374)
• Add utilities for ANSI parsing, caching styles, and diff generation (a20742ed)
• Update configuration for agent provider and default model; adjust theme settings (465ed6e7)
• Enhance changelog generation with Highlights and Contributors sections (8e25479e)
• Enhance tool output handling and caching logic (8635f4a0)
• Refactor unified_exec tool to add inspect action and improve documentation (eac81276)
• Refactor guidance messages for clarity and conciseness; update tool output handling and error messages. (ae454f4b)
• Add 'continue' action to unified_exec tool; refactor related logic and documentation (b2102ff4)
• Refactor tool policies and enhance shell command parsing; add next_continue_args handling (327a2467)
• Add clear screen and copy latest assistant reply commands; update related documentation and tests (39b2d94a)
• Enhance command detection by normalizing natural language requests and adding support for various command patterns (bc2be0b9)
• Change terminal theme from 'ciapre-dark' to 'mono' (6b03347c)
• Add appearance configuration and improve modal handling (afd21755)
• Enhance external editor integration and rewind functionality (e5f722ce)
• Refactor code for improved readability and functionality; normalize unified search arguments; enhance validation error handling; update default model in configuration. (56adb18f)
• Refactor palette handling and LLM retry logic; improve error handling and tests (64f0f949)
• Enhance task tracker functionality and error handling (42781438)
• Remove XAI provider tests and related assertions from model validation (880ec280)
• Remove deprecated integration and workflow tests; clean up test module imports (3c9f5f42)
• Update model references from "gemini-3-pro-preview" to "gemini-3.1-pro-preview" across the codebase (468a0765)
• Refactor caching mechanisms to use Mutex instead of RwLock for improved concurrency (af7ebf32)
• Refactor sandbox configuration and enhance ANSI capabilities (cfb2f51a)
• Update default model to GPT-5.3 Codex and enhance terminal title management (9d3d7d38)
• Update default model to GPT-5.3 Codex, change provider to OpenAI, and enhance reasoning effort configuration (24d3af83)
• Remove xAI provider support and update default models in configuration files (ad6290aa)
• Refactor code for improved readability and consistency in function calls (91fdb43a)

Refactors

• Streamline task tracking logic and introduce TaskTrackingStatus for improved status management (3be2c784)

Contributors

@vinhnx

Full Changelog: https://github.com/vinhnx/vtcode/compare/0.83.0...0.84.0