v0.1.0

v0.1.0 - Initial Release

Features

• Image interception via experimental.chat.messages.transform hook
• Local VL API support (OpenAI-compatible, default: Qwen3VL via llama.cpp)
• Multi-image support with deduplication and numbered labels
• Smart skip: auto-detects when the current chat model natively supports images
• visionModels config for pattern-based vision-capable model matching
• HTTP URL and data URL image support
• JSONC config loading (project-level + user-level + built-in defaults)
• Debug logging
• CI/CD: GitHub Actions for syntax check and npm publish
• Multilingual README: 10 languages

Installation

// opencode.json
{ "plugin": ["opencode-vision-paste"] }