Bring in mcp-skills-plugins branch: marketplace, plugins, skills, OpenCredits, and image previews

Major release adding:
- MCP marketplace with curated registry and search across multiple registries
- Plugins and skills marketplace integration
- OpenCredits payment integration with model selection and checkout flow
- Image preview before sending (paste, file picker)
- Self-hosted Umami analytics with custom events
- Support feedback modal with Discord webhook
- Local OpenAI to Anthropic router for model routing
- Inline stop button replacing send during processing
- Improved install flow requiring Node.js 18+
- WSL env var passthrough fixes
- Better error handling and Windows compatibility

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

src/recommended-models.json

@@ -0,0 +1,66 @@
+[
+    {
+        "id": "openai/gpt-5.3-codex",
+        "name": "GPT 5.3 Codex",
+        "description": "Coding-focused GPT-5.3 variant with optimized routing.",
+        "context_length": 400000,
+        "max_output_tokens": 128000,
+        "credits_per_request": 4.921875,
+        "provider": "OpenAI",
+        "quickLabel": "GPT",
+        "tierModels": { "sonnet": "openai/gpt-5.3-codex", "opus": "openai/gpt-5.3-codex", "haiku": "openai/gpt-5.1-codex-mini" }
+    },
+    {
+        "id": "google/gemini-3.1-pro-preview",
+        "name": "Gemini 3.1 Pro Preview",
+        "description": "Google's Gemini 3.1 Pro with enhanced reasoning and multimodal support.",
+        "context_length": 1000000,
+        "max_output_tokens": 64000,
+        "credits_per_request": 4.375,
+        "provider": "Google",
+        "quickLabel": "Gemini",
+        "tierModels": { "sonnet": "google/gemini-3.1-pro-preview", "opus": "google/gemini-3.1-pro-preview", "haiku": "google/gemini-3-flash" }
+    },
+    {
+        "id": "minimax/minimax-m2.7",
+        "name": "Minimax M2.7",
+        "description": "MiniMax M2.7 with enhanced context understanding and improved complex tool use. Optimized for agentic workflows and long-horizon tasks.",
+        "context_length": 204800,
+        "max_output_tokens": 131000,
+        "credits_per_request": 0.46875,
+        "provider": "MiniMax",
+        "quickLabel": "MiniMax"
+    },
+    {
+        "id": "moonshotai/kimi-k2.5",
+        "name": "Kimi K2.5",
+        "description": "Kimi K2.5 is Moonshot AI's native multimodal model with strong general reasoning, visual coding, and agentic tool-calling.",
+        "context_length": 262114,
+        "max_output_tokens": 262114,
+        "credits_per_request": 1.125,
+        "provider": "Moonshot AI",
+        "quickLabel": "Kimi",
+        "tierModels": { "sonnet": "moonshotai/kimi-k2.5", "opus": "moonshotai/kimi-k2.5", "haiku": "moonshotai/kimi-k2-turbo" }
+    },
+    {
+        "id": "zai/glm-5",
+        "name": "GLM 5",
+        "description": "GLM-5 is the latest GLM series text model with stronger reasoning, long-context chat, and reliable tool use.",
+        "context_length": 202800,
+        "max_output_tokens": 131100,
+        "credits_per_request": 1.3125,
+        "provider": "Zhipu AI",
+        "quickLabel": "GLM",
+        "tierModels": { "sonnet": "zai/glm-5", "opus": "zai/glm-5", "haiku": "zai/glm-4.7-flash" }
+    },
+    {
+        "id": "deepseek/deepseek-v3.2-thinking",
+        "name": "DeepSeek V3.2 Thinking",
+        "description": "DeepSeek V3.2 thinking/reasoner mode. Reasoning-first model built for agents. First DeepSeek model with thinking-in-tool-use capability.",
+        "context_length": 128000,
+        "max_output_tokens": 64000,
+        "credits_per_request": 0.21875,
+        "provider": "DeepSeek",
+        "quickLabel": "DeepSeek"
+    }
+]