Changelog

All notable changes to OpenConduit will be documented here.

1.2.1 (2026-05-18)

Add native Ollama provider with localhost detection and dynamic model discovery (3936377)
artifacts rendering + saved files panel (9091e96)
artifacts rendering + saved files panel (#9) (a327a95)
core: add setConversations, personaId/workspaceId to Conversation type (4523286)
core: multi-model comparison mode with multi-turn history (#11) (a9c3549)
core: multi-model comparison mode with multi-turn history (#11) (504901f)
core: SettingsPanel accepts extraTabs and hideTabs props (f01c14f)
desktop: add Google Gemini provider (#6) (e217170)
desktop: add Google Gemini provider (#6) (c8d8a69)
file & image attachment support (#5) (f84bae9)
file & image attachment support (#5) (e61c37f)
intelligent model routing with profiles, compare picker, and streaming fixes (2de44f5)
intelligent model routing with profiles, compare picker, and streaming fixes (46184a0)
MCP server + provider marketplace (#10, #31) (01315b7)
MCP server + provider marketplace with per-conversation tool toggles (941d99a)

add vite/client types to tsconfig for import.meta.glob support (207d0d8)
beta channel shows only beta builds, alpha channel shows alpha+beta (a0697cf)
class-based dark mode for Tailwind v4 and system theme listener (c257503), closes #19
resolve @openconduit/core as npm package in Vite (665a0d5)
resolve @openconduit/core as npm package in Vite (951cd93)
theme switching (#19) (a5930d6)

Multi-provider AI chat: OpenAI, Anthropic (Claude), and LM Studio (OpenAI-compatible)
Streaming responses with real-time token output
Persistent conversation history with sidebar navigation
Per-conversation system prompt editor
MCP (Model Context Protocol) tool server support — HTTP-SSE and stdio transports
Tool call approval UI — allow or deny individual tool invocations
AI Task Tracking (Labs) — AI maintains a live task list during multi-step work
AI Clarifying Questions (Labs) — AI asks targeted questions inline before proceeding
Parameter controls — temperature, max tokens, top-p per conversation
Context window usage indicator with overflow warning
Conversation export as JSON or Markdown
Analytics tab — token usage and cost tracking across conversations
Update checker — polls Cloudflare Worker for latest GitHub release
In-app feedback — submit bug reports and feature requests as GitHub Issues
Dark mode support (system, light, dark)
macOS native window with hidden-inset title bar and traffic-light spacing
Cross-platform packaging for macOS (arm64 + x64), Windows, and Linux via electron-forge