Rebrand GoVectorSync to Distill throughout codebase

- Update CLI name from govs to distill
- Update all help text and examples
- Update MCP server name and resources (distill://)
- Update config file name (.distill.yaml)
- Update mcp/ directory docs and examples
- Consistent branding as 'Distill' everywhere

Author: Siddhant-K-code

cmd/analyze.go

@@ -23,7 +23,7 @@ var analyzeCmd = &cobra.Command{
 potential duplicates without uploading to Pinecone.
 
 Example:
-  govs analyze --file data.jsonl --threshold 0.05
+  distill analyze --file data.jsonl --threshold 0.05
 
 The threshold controls duplicate sensitivity:
   - 0.01: Very strict (only near-identical vectors)
@@ -183,7 +183,7 @@ func printAnalysisReport(result *types.DeduplicationResult, verbose bool) {
 	fmt.Println()
 
 	if result.DuplicateCount > 0 {
-		fmt.Println("Recommendation: Use 'govs sync --dedup=true' to upload deduplicated vectors.")
+		fmt.Println("Recommendation: Use 'distill sync --dedup=true' to upload deduplicated vectors.")
 	} else {
 		fmt.Println("No duplicates found. Your dataset is already unique.")
 	}

cmd/api.go

@@ -25,7 +25,7 @@ Unlike 'serve', this doesn't require a vector DB connection.
 Clients send chunks directly and receive deduplicated results.
 
 Example:
-  govs api --port 8080
+  distill api --port 8080
 
 The server exposes:
   POST /v1/dedupe   - Deduplicate chunks

cmd/mcp.go

@@ -20,10 +20,10 @@ import (
 
 var mcpCmd = &cobra.Command{
 	Use:   "mcp",
-	Short: "Start GoVectorSync as an MCP server",
-	Long: `Starts GoVectorSync as a Model Context Protocol (MCP) server.
+	Short: "Start Distill as an MCP server",
+	Long: `Starts Distill as a Model Context Protocol (MCP) server.
 
-This allows AI assistants like Claude, Amp, and Cursor to use GoVectorSync's
+This allows AI assistants like Claude, Amp, and Cursor to use Distill's
 semantic deduplication capabilities directly.
 
 Transports:
@@ -36,23 +36,23 @@ Tools exposed:
   analyze_redundancy    - Analyze chunks for redundancy stats
 
 Resources exposed:
-  govectorsync://system-prompt - System prompt for AI assistants
+  distill://system-prompt - System prompt for AI assistants
 
 Example:
   # Local stdio server (Claude Desktop, Cursor, Amp)
-  govs mcp
+  distill mcp
 
   # Remote HTTP server (hosted deployment)
-  govs mcp --transport http --port 8081
+  distill mcp --transport http --port 8081
 
   # With vector DB backend
-  govs mcp --backend pinecone --index my-index
+  distill mcp --backend pinecone --index my-index
 
 Configure in Claude Desktop (claude_desktop_config.json):
   {
     "mcpServers": {
-      "govectorsync": {
-        "command": "govs",
+      "distill": {
+        "command": "distill",
         "args": ["mcp"]
       }
     }
@@ -61,7 +61,7 @@ Configure in Claude Desktop (claude_desktop_config.json):
 For remote MCP server:
   {
     "mcpServers": {
-      "govectorsync": {
+      "distill": {
         "url": "https://your-server.fly.dev/mcp"
       }
     }
@@ -95,7 +95,7 @@ func init() {
 	mcpCmd.Flags().Float64("lambda", 0.5, "Default MMR lambda")
 }
 
-// MCPServer wraps the MCP server with GoVectorSync capabilities
+// MCPServer wraps the MCP server with Distill capabilities
 type MCPServer struct {
 	broker   *contextlab.Broker
 	embedder retriever.EmbeddingProvider
@@ -209,7 +209,7 @@ func runMCP(cmd *cobra.Command, args []string) error {
 
 	// Create MCP server with capabilities
 	s := server.NewMCPServer(
-		"GoVectorSync",
+		"Distill",
 		"1.0.0",
 		server.WithToolCapabilities(false),
 		server.WithResourceCapabilities(true, false),
@@ -230,7 +230,7 @@ func runMCP(cmd *cobra.Command, args []string) error {
 
 	case "http":
 		addr := fmt.Sprintf("%s:%d", host, port)
-		fmt.Printf("GoVectorSync MCP server starting on http://%s\n", addr)
+		fmt.Printf("Distill MCP server starting on http://%s\n", addr)
 		fmt.Printf("  Endpoint: http://%s/mcp\n", addr)
 		fmt.Printf("  Health:   http://%s/health\n", addr)
 		fmt.Println()
@@ -241,7 +241,7 @@ func runMCP(cmd *cobra.Command, args []string) error {
 		// Health check endpoint
 		mux.HandleFunc("/health", func(w http.ResponseWriter, r *http.Request) {
 			w.Header().Set("Content-Type", "application/json")
-			_, _ = w.Write([]byte(`{"status":"ok","server":"govectorsync-mcp"}`))
+			_, _ = w.Write([]byte(`{"status":"ok","server":"distill-mcp"}`))
 		})
 
 		// MCP endpoint with stateful sessions
@@ -350,7 +350,7 @@ whether to deduplicate. Returns cluster information and redundancy percentage.`)
 }
 
 // System prompt that guides AI assistants to use deduplication
-const systemPromptContent = `You have access to GoVectorSync, a semantic deduplication tool for RAG context.
+const systemPromptContent = `You have access to Distill, a semantic deduplication tool for RAG context.
 
 IMPORTANT: When you receive multiple chunks from retrieval or search:
 1. Check if chunks might contain overlapping information
@@ -366,16 +366,16 @@ Use analyze_redundancy first if you want to see the redundancy stats before dedu
 func (m *MCPServer) registerResources(s *server.MCPServer) {
 	// System prompt resource - hosts can include this in context
 	systemPrompt := mcp.NewResource(
-		"govectorsync://system-prompt",
-		"GoVectorSync System Prompt",
+		"distill://system-prompt",
+		"Distill System Prompt",
 		mcp.WithResourceDescription("System prompt that guides AI to use deduplication tools effectively"),
 		mcp.WithMIMEType("text/plain"),
 	)
 
 	s.AddResource(systemPrompt, func(ctx context.Context, request mcp.ReadResourceRequest) ([]mcp.ResourceContents, error) {
 		return []mcp.ResourceContents{
 			mcp.TextResourceContents{
-				URI:      "govectorsync://system-prompt",
+				URI:      "distill://system-prompt",
 				MIMEType: "text/plain",
 				Text:     systemPromptContent,
 			},
@@ -384,8 +384,8 @@ func (m *MCPServer) registerResources(s *server.MCPServer) {
 
 	// Configuration resource - shows current settings
 	configResource := mcp.NewResource(
-		"govectorsync://config",
-		"GoVectorSync Configuration",
+		"distill://config",
+		"Distill Configuration",
 		mcp.WithResourceDescription("Current deduplication configuration and defaults"),
 		mcp.WithMIMEType("application/json"),
 	)
@@ -404,7 +404,7 @@ func (m *MCPServer) registerResources(s *server.MCPServer) {
 		configJSON, _ := json.MarshalIndent(config, "", "  ")
 		return []mcp.ResourceContents{
 			mcp.TextResourceContents{
-				URI:      "govectorsync://config",
+				URI:      "distill://config",
 				MIMEType: "application/json",
 				Text:     string(configJSON),
 			},

cmd/query.go

@@ -25,7 +25,7 @@ var queryCmd = &cobra.Command{
 Useful for testing and tuning ContextLab parameters.
 
 Example:
-  govs query "How do I configure authentication?" --index my-index
+  distill query "How do I configure authentication?" --index my-index
 
 Requires PINECONE_API_KEY and OPENAI_API_KEY environment variables.`,
 	Args: cobra.MinimumNArgs(1),

cmd/root.go

@@ -12,21 +12,21 @@ var cfgFile string
 
 // rootCmd represents the base command when called without any subcommands
 var rootCmd = &cobra.Command{
-	Use:   "govs",
-	Short: "GoVectorSync - High-performance vector ingestion with semantic deduplication",
-	Long: `GoVectorSync (govs) is a production-grade CLI tool for ingesting massive 
-datasets of vector embeddings into Pinecone with client-side semantic deduplication.
+	Use:   "distill",
+	Short: "Distill - Reliability layer for LLM context with semantic deduplication",
+	Long: `Distill is a reliability layer for LLM context that removes redundancy
+before it reaches your model, improving output quality and determinism.
 
 Features:
-  - SIMD-accelerated cosine distance calculations (AVX2/AVX-512)
-  - Custom K-Means clustering for semantic deduplication
-  - Worker pool pattern for maximum throughput
-  - Exponential backoff retry logic
+  - Agglomerative clustering for semantic deduplication
+  - MMR re-ranking for diversity
+  - ~12ms latency, no LLM calls
+  - Deterministic, auditable results
 
 Environment Variables:
-  PINECONE_API_KEY    Your Pinecone API key (required for sync)
-  PINECONE_INDEX      Default index name
-  PINECONE_NAMESPACE  Default namespace`,
+  OPENAI_API_KEY      For text → embedding conversion
+  PINECONE_API_KEY    For Pinecone backend
+  QDRANT_URL          For Qdrant backend`,
 }
 
 // Execute adds all child commands to the root command and sets flags appropriately.
@@ -41,7 +41,7 @@ func init() {
 	cobra.OnInitialize(initConfig)
 
 	// Global flags
-	rootCmd.PersistentFlags().StringVar(&cfgFile, "config", "", "config file (default is $HOME/.govs.yaml)")
+	rootCmd.PersistentFlags().StringVar(&cfgFile, "config", "", "config file (default is $HOME/.distill.yaml)")
 	rootCmd.PersistentFlags().Bool("verbose", false, "enable verbose output")
 
 	// Bind to viper
@@ -59,7 +59,7 @@ func initConfig() {
 		}
 		viper.AddConfigPath(".")
 		viper.SetConfigType("yaml")
-		viper.SetConfigName(".govs")
+		viper.SetConfigName(".distill")
 	}
 
 	// Read environment variables

cmd/serve.go

@@ -27,7 +27,7 @@ var serveCmd = &cobra.Command{
 for RAG retrieval queries.
 
 Example:
-  govs serve --port 8080 --backend pinecone --index my-index
+  distill serve --port 8080 --backend pinecone --index my-index
 
 The server exposes:
   POST /v1/retrieve  - Deduplicated retrieval endpoint

cmd/sync.go

@@ -23,7 +23,7 @@ var syncCmd = &cobra.Command{
 and uploads to a Pinecone index using parallel workers.
 
 Example:
-  govs sync --file data.jsonl --index my-index --dedup=true
+  distill sync --file data.jsonl --index my-index --dedup=true
 
 Environment Variables:
   PINECONE_API_KEY    Your Pinecone API key (required)`,

mcp/README.md

@@ -1,17 +1,17 @@
-# GoVectorSync MCP Integration
+# Distill MCP Integration
 
-GoVectorSync exposes semantic deduplication as an MCP (Model Context Protocol) server, allowing AI assistants to deduplicate context directly.
+Distill exposes semantic deduplication as an MCP (Model Context Protocol) server, allowing AI assistants to deduplicate context directly.
 
 ## The Problem
 
 When AI assistants assemble context from multiple sources (code, docs, memory, tool outputs), 30-40% is typically redundant. Same information from different sources wastes tokens and confuses the model.
 
-## How GoVectorSync Helps
+## How Distill Helps
 
-GoVectorSync sits between your context sources and the LLM:
+Distill sits between your context sources and the LLM:
 
 ```
-RAG Results → GoVectorSync → Deduplicated Context → LLM
+RAG Results → Distill → Deduplicated Context → LLM
 ```
 
 It clusters semantically similar chunks, picks the best representative from each cluster, and applies MMR for diversity.
@@ -22,17 +22,17 @@ It clusters semantically similar chunks, picks the best representative from each
 
 ```bash
 # Build
-go build -o govs .
+go build -o distill .
 
 # Start MCP server
-./govs mcp
+./distill mcp
 ```
 
 ### Remote (HTTP) - Hosted deployment
 
 ```bash
 # Start HTTP server
-./govs mcp --transport http --port 8081
+./distill mcp --transport http --port 8081
 
 # Or deploy to Fly.io
 fly deploy -c fly.mcp.toml
@@ -76,11 +76,11 @@ Analyze chunks for redundancy without removing any. Use to understand overlap be
 
 ## Resources
 
-### `govectorsync://system-prompt`
+### `distill://system-prompt`
 
 System prompt that guides AI assistants to use deduplication effectively. Host applications can include this in context automatically.
 
-### `govectorsync://config`
+### `distill://config`
 
 Current configuration and defaults (JSON).
 
@@ -106,8 +106,8 @@ Add to `~/Library/Application Support/Claude/claude_desktop_config.json`:
 ```json
 {
   "mcpServers": {
-    "govectorsync": {
-      "command": "/path/to/govs",
+    "distill": {
+      "command": "/path/to/distill",
       "args": ["mcp"]
     }
   }
@@ -118,8 +118,8 @@ Add to `~/Library/Application Support/Claude/claude_desktop_config.json`:
 ```json
 {
   "mcpServers": {
-    "govectorsync": {
-      "url": "https://govectorsync-mcp.fly.dev/mcp"
+    "distill": {
+      "url": "https://distill-mcp.fly.dev/mcp"
     }
   }
 }
@@ -129,8 +129,8 @@ Add to `~/Library/Application Support/Claude/claude_desktop_config.json`:
 ```json
 {
   "mcpServers": {
-    "govectorsync": {
-      "command": "/path/to/govs",
+    "distill": {
+      "command": "/path/to/distill",
       "args": ["mcp", "--backend", "pinecone", "--index", "my-index"],
       "env": {
         "PINECONE_API_KEY": "your-api-key",
@@ -165,25 +165,25 @@ fly secrets set PINECONE_API_KEY=xxx -c fly.mcp.toml
 fly deploy -c fly.mcp.toml
 ```
 
-Your MCP endpoint: `https://govectorsync-mcp.fly.dev/mcp`
+Your MCP endpoint: `https://distill-mcp.fly.dev/mcp`
 
 ### Docker
 
 ```bash
 # Build
-docker build -f Dockerfile.mcp -t govectorsync-mcp .
+docker build -f Dockerfile.mcp -t distill-mcp .
 
 # Run
 docker run -p 8081:8081 \
   -e OPENAI_API_KEY=xxx \
-  govectorsync-mcp
+  distill-mcp
 ```
 
 ### Railway / Render / Other
 
 Use the Dockerfile.mcp and set:
 - Port: 8081
-- Command: `./govs mcp --transport http --port 8081`
+- Command: `./distill mcp --transport http --port 8081`
 
 ## Integration Patterns
 
@@ -192,7 +192,7 @@ Use the Dockerfile.mcp and set:
 The AI reads the system prompt resource and automatically deduplicates when it detects multiple chunks:
 
 ```
-1. Host app includes govectorsync://system-prompt in context
+1. Host app includes distill://system-prompt in context
 2. User asks a question
 3. RAG retrieves chunks
 4. AI recognizes overlap, calls deduplicate_chunks
@@ -234,7 +234,7 @@ AI: "Found 8 diverse results from 50 retrieved"
 The AI decides to call tools based on:
 
 1. **Tool descriptions** - We've written action-oriented descriptions that explain when to use each tool
-2. **System prompt** - The `govectorsync://system-prompt` resource guides the AI to check for redundancy
+2. **System prompt** - The `distill://system-prompt` resource guides the AI to check for redundancy
 3. **User requests** - Explicit requests like "deduplicate these" or "remove redundancy"
 4. **Context patterns** - When the AI sees multiple similar chunks, it may recognize the need
 
@@ -260,8 +260,8 @@ The AI decides to call tools based on:
 
 ## Comparison: MCP vs HTTP Proxy
 
-| Aspect | MCP Server | HTTP Proxy (`govs serve`) |
-|--------|------------|---------------------------|
+| Aspect | MCP Server | HTTP Proxy (`distill serve`) |
+|--------|------------|------------------------------|
 | Invocation | AI decides when to call | Automatic on every query |
 | Integration | MCP-compatible clients | Any HTTP client |
 | Use case | AI assistant workflows | RAG pipeline middleware |