added z.ai, gemini

.claude/settings.local.json

@@ -9,7 +9,25 @@
       "Bash(npm run test:unit:*)",
       "Bash(node -e:*)",
       "Bash(npm test:*)",
-      "Bash(npm rebuild:*)"
+      "Bash(npm rebuild:*)",
+      "WebFetch(domain:arxiv.org)",
+      "WebFetch(domain:llmlingua.com)",
+      "WebFetch(domain:pypi.org)",
+      "WebFetch(domain:api.python.langchain.com)",
+      "WebFetch(domain:microsoft.github.io)",
+      "WebFetch(domain:reference.langchain.com)",
+      "Bash(gh repo view:*)",
+      "Bash(gh api:*)",
+      "Bash(npm start)",
+      "Bash(curl:*)",
+      "Bash(lsof:*)",
+      "Bash(tail:*)",
+      "Bash(pkill:*)",
+      "Bash(jq:*)",
+      "Bash(jobs:*)",
+      "Bash(gh pr view:*)",
+      "Bash(gh run list:*)",
+      "Bash(gh run view:*)"
     ],
     "deny": [],
     "ask": []

.env.example

@@ -6,7 +6,7 @@
 # ==============================================================================
 
 # Primary model provider to use
-# Options: databricks, azure-anthropic, azure-openai, openrouter, openai, ollama, llamacpp, lmstudio, bedrock
+# Options: databricks, azure-anthropic, azure-openai, openrouter, openai, ollama, llamacpp, lmstudio, bedrock, zai, vertex
 # Default: databricks
 MODEL_PROVIDER=ollama
 
@@ -143,6 +143,44 @@ OLLAMA_MAX_TOOLS_FOR_ROUTING=3
 # LMSTUDIO_TIMEOUT_MS=120000
 # LMSTUDIO_API_KEY=your-optional-api-key
 
+# ==============================================================================
+# Z.AI (Zhipu AI) Configuration - ~1/7 cost of Anthropic
+# ==============================================================================
+
+# Z.AI provides GLM models through an Anthropic-compatible API
+# Get your API key from: https://z.ai/ or https://open.bigmodel.cn/
+# ZAI_API_KEY=your-zai-api-key
+
+# Z.AI endpoint (default: https://api.z.ai/api/anthropic/v1/messages)
+# ZAI_ENDPOINT=https://api.z.ai/api/anthropic/v1/messages
+
+# Model to use (GLM-4.7 is equivalent to Claude Sonnet, GLM-4.5-Air is like Haiku)
+# Options: GLM-4.7, GLM-4.5-Air, GLM-4-Plus
+# ZAI_MODEL=GLM-4.7
+
+# ==============================================================================
+# Google Vertex AI Configuration (Gemini Models)
+# ==============================================================================
+
+# Google AI API Key (required)
+# Get your API key from: https://aistudio.google.com/app/apikey
+# VERTEX_API_KEY=your-google-api-key
+# or use: GOOGLE_API_KEY=your-google-api-key
+
+# Gemini model to use (default: gemini-2.0-flash)
+# Options:
+#   - gemini-2.0-flash (fast, good for most tasks) - DEFAULT
+#   - gemini-2.0-flash-lite (fastest, cheapest)
+#   - gemini-2.5-pro (most capable, best quality)
+#   - gemini-1.5-pro (previous generation)
+#   - gemini-1.5-flash (previous generation, fast)
+# VERTEX_MODEL=gemini-2.0-flash
+
+# Model mapping from Claude names:
+#   claude-sonnet-* → gemini-2.0-flash
+#   claude-haiku-*  → gemini-2.0-flash-lite
+#   claude-opus-*   → gemini-2.5-pro
+
 # ==============================================================================
 # Embeddings Provider Override
 # ==============================================================================
@@ -253,6 +291,17 @@ API_RETRY_MAX_DELAY=30000
 LOAD_SHEDDING_HEAP_THRESHOLD=90
 LOAD_SHEDDING_EVENT_LOOP_DELAY=100
 
+# ==============================================================================
+# Hot Reload Configuration
+# ==============================================================================
+
+# Enable hot reload of configuration (default: true)
+# When enabled, changes to .env are applied without restart
+HOT_RELOAD_ENABLED=true
+
+# Debounce delay in ms (prevents rapid reloads)
+HOT_RELOAD_DEBOUNCE_MS=1000
+
 # ==============================================================================
 # Quick Start Examples
 # ==============================================================================
@@ -279,3 +328,15 @@ LOAD_SHEDDING_EVENT_LOOP_DELAY=100
 #   AZURE_OPENAI_API_KEY=your-key
 #   AZURE_OPENAI_DEPLOYMENT=gpt-4o
 #   npm start
+
+# Z.AI (Zhipu - ~1/7 cost of Anthropic):
+#   MODEL_PROVIDER=zai
+#   ZAI_API_KEY=your-zai-api-key
+#   ZAI_MODEL=GLM-4.7
+#   npm start
+
+# Google Gemini (via Vertex AI):
+#   MODEL_PROVIDER=vertex
+#   VERTEX_API_KEY=your-google-api-key
+#   VERTEX_MODEL=gemini-2.0-flash
+#   npm start

.gitignore

@@ -21,3 +21,5 @@ ROUTER_COMPARISON.md
 TIER_ROUTING_PLAN.md
 docs/GSD_LEARNINGS.md
 LYNKR-TUI-PLAN.md
+HEADROOM_PORT_PLAN.md
+HEADROOM_SIDECAR_PLAN.md

package.json

@@ -1,6 +1,6 @@
 {
   "name": "lynkr",
-  "version": "4.2.1",
+  "version": "4.3.1",
   "description": "Self-hosted Claude Code & Cursor proxy with Databricks,AWS BedRock,Azure  adapters, openrouter, Ollama,llamacpp,LM Studio, workspace tooling, and MCP integration.",
   "main": "index.js",
   "bin": {