v1.2.6: New D/Clojure Code Formatting, GitLab Improvements
Release Date: February 16, 2026
One-Sentence Summary
This update adds dfmt and cljfmt code formatters for D and Clojure, and improves GitLab login experience.
New Feature: D Language Formatting (dfmt)
What was updated?
Added dfmt formatter support for D language files.
What is dfmt?
dfmt is the official code formatting tool for D language, which automatically adjusts code indentation, spacing, line breaks, etc., for consistent code style.
How to use?
When you edit .d files, OpenCode will automatically use dfmt to format the code.
If you want automatic code formatting, you can enable it in the configuration file:
{
"formatters": {
"d": "dfmt"
}
}What is D language?
D language is a systems programming language that combines C++ performance with Python usability. It's used for:
- High-performance applications
- Game development
- Systems programming
New Feature: Clojure Formatting (cljfmt)
What was updated?
Added cljfmt formatter support for Clojure files.
What is cljfmt?
cljfmt is a code formatting tool for Clojure, based on the core.style style guide.
How to use?
When editing .clj or .cljs files, OpenCode will automatically use cljfmt for formatting.
{
"formatters": {
"clojure": "cljfmt"
}
}What is Clojure?
Clojure is a Lisp dialect running on the JVM platform, known for simplicity and functional programming. It's used for:
- Data processing
- Web development
- Concurrent programming
New Feature: GitLab Mid-Session Token Refresh
What was updated?
Upgraded GitLab provider and authentication plugin, now supporting mid-session token refresh without needing to re-login.
What are the benefits?
Previously, if your GitLab token expired, you needed to re-login. Now:
- Token automatically refreshes when about to expire
- Won't interrupt your work
- Smoother experience
When does it refresh?
- When token is about to expire (usually hourly)
- Won't affect ongoing conversations
New Feature: Google Vertex OpenAI Compatible Endpoint
What was updated?
Added OpenAI compatible endpoint support for the Google Vertex provider.
What are the benefits?
If you have tools or services that support the OpenAI API format, you can now use them to access Google Vertex models.
Simply put: More flexible model access methods.
New Feature: Venice Parameter Support
What was updated?
Added more parameter support for the Venice provider:
- temperature
- topP
- topK
- smallOption
What are the benefits?
More fine-grained control over Venice model output:
- temperature: Control creativity level (higher = more random)
- topP: Nucleus sampling parameter
- topK: Limit candidate word count
- smallOption: Small model option
Bug Fix: Environment Variable Priority
The Previous Problem
In Google Vertex AI configuration, environment variables might override user configuration, causing settings not to take effect.
The Current Situation
Now user-defined variables take priority over environment variables, so your configuration won't be overridden.
Bug Fix: Unnecessary LLM Calls
The Previous Problem
Every message would call LLM to generate a title, causing unnecessary API calls and latency.
The Current Situation
Removed unnecessary per-message title LLM calls, improving response speed.
New Feature: TUI Database Migration Command
What was updated?
Added a database migration command in TUI mode to migrate JSON storage to SQLite.
How to use?
Enter the /migrate command in the TUI interface to trigger migration.
Note: Migration may take some time depending on data volume.
New Feature: attach Command Extensions
What was updated?
Added --continue and --fork flags to the attach command.
How to use?
opencode attach --continue: Continue previous sessionopencode attach --fork: Copy current session and create new branch
Bug Fix: Desktop Linux Display
The Previous Problem
When using the desktop application on Linux, Wayland/X11 backend and decoration strategy were inconsistent.
The Current Situation
Unified Linux display backend and window decoration strategy for more consistent experience.
Should You Upgrade?
| Situation | Recommendation |
|---|---|
| Using D or Clojure | ✅ Upgrade |
| Using GitLab | ✅ Upgrade |
| Desktop Linux user | ✅ Upgrade |
| Want faster response | ✅ Upgrade |
| Everything works fine | Can wait for next release |
Upgrade command:
opencode upgradeTechnical Details (Optional)
Click to expand full changelog
Core
- Added dfmt formatter support for D language files
- Upgraded GitLab provider and authentication plugin to support mid-session token refresh
- Removed unnecessary per-message title LLM calls
- User-defined variables take priority over environment variables in Google Vertex AI configuration
- Added OpenAI compatible endpoint support for Google Vertex provider
- Added temperature, topP, topK, and smallOption parameter support for Venice
- Added cljfmt formatter support for Clojure files
TUI (Terminal Interface)
- Use server directory path to handle file references in prompts
- Added database migration command to convert JSON storage to SQLite
- Added --continue and --fork flags to attach command
- Fixed inaccurate prompts in TUI
Desktop
- Unified Linux Wayland/X11 backend and decoration strategy
- Use process-wrap library in desktop application instead of manual job object handling
Community Contributors
Thanks to 12 community contributors:
- @finalfantasia: Added cljfmt formatter
- @pkx07: Fixed proper nouns in Chinese translation
- @Brendonovich: Use process-wrap instead of manual job objects
- @dpuyosa: Added parameter support for Venice
- @bnema: Unified Linux display backend
- @leehack: Added OpenAI compatible endpoint for google-vertex
- @hobostay: Fixed Russian documentation translation errors
- @rmk40: Removed unnecessary LLM title calls
- @imanolmzd-svg: Fixed inaccurate prompts
- @vglafirov: Upgraded GitLab provider
- @OpeOginni: Fixed file reference path issue
- @burner: Added dfmt formatter