feat: add TokenTracker for managing token usage, costs, and context.

Commands/SetupCommand.cs

@@ -0,0 +1,13 @@
+namespace AnchorCli.Commands;
+
+public class SetupCommand : ICommand
+{
+    public string Name => "setup";
+    public string Description => "Run interactive setup wizard to configure API key and model";
+
+    public Task ExecuteAsync(string[] args, CancellationToken ct)
+    {
+        SetupTui.Run();
+        return Task.CompletedTask;
+    }
+}

OpenRouter/TokenTracker.cs

@@ -37,15 +37,15 @@ internal sealed class TokenTracker
 
     /// <summary>
     /// Returns true if the context is getting too large and should be compacted.
-    /// Triggers at min(75% of model context, 100K tokens).
+    /// Triggers at min(75% of model context, 150K tokens).
     /// </summary>
     public bool ShouldCompact()
     {
         if (LastInputTokens <= 0) return false;
 
         int threshold = ContextLength > 0
-            ? Math.Min((int)(ContextLength * 0.75), 100_000)
+            ? Math.Min((int)(ContextLength * 0.75), 150_000)
-            : 100_000;
+            : 150_000;
 
         return LastInputTokens >= threshold;
     }

Program.cs

@@ -183,6 +183,7 @@ commandRegistry.Register(new HelpCommand(commandRegistry));
 commandRegistry.Register(new ClearCommand());
 commandRegistry.Register(new StatusCommand(model, endpoint));
 commandRegistry.Register(new CompactCommand(compactor, history));
+commandRegistry.Register(new SetupCommand());
 
 var commandDispatcher = new CommandDispatcher(commandRegistry);
 

README.md

@@ -24,7 +24,8 @@ This eliminates:
 - **Interactive REPL**: Chat with an AI model to edit files, manage directories, and execute commands
 - **Slash Commands**: `/setup`, `/help`, `/exit`, `/clear`, `/status`, `/compact`
 - **Token Tracking**: Real-time token usage and cost per response, plus session totals
-- **Context Compaction**: Automatic conversation history compression when approaching context limits
+- **Model Pricing Display**: Shows current model pricing from OpenRouter in the header
+- **Context Compaction**: Automatic conversation history compression when approaching context limits, including stale tool result compaction
 - **Comprehensive Toolset**: 15 tools for file operations, editing, directory management, and command execution
 - **AOT-Ready**: Native AOT compilation for ~12 MB binaries with no .NET runtime dependency
 - **Rich CLI**: Beautiful terminal output using Spectre.Console with tables, rules, and colored text
@@ -40,13 +41,11 @@ This eliminates:
 ## Quick Start
 
 ```bash
-# Option 1: Set environment variables
+# Run the application
-export ANCHOR_API_KEY=your_key_here
-export ANCHOR_MODEL=qwen3.5-27b       # optional, default: gpt-4o
-
 dotnet run --project AnchorCli
 
-# Option 2: Use interactive setup
+# First time? The app will prompt you to run /setup
+# Or run it explicitly:
 dotnet run --project AnchorCli
 /setup
 ```
@@ -60,20 +59,11 @@ dotnet publish AnchorCli -r linux-x64 -c Release
 
 The resulting binary is ~12 MB, has no .NET runtime dependency, and starts instantly.
 
-## Environment Variables
-
-| Variable | Default | Description |
-|---|---|---|
-| `ANCHOR_API_KEY` | *(required)* | API key for the LLM provider |
-| `ANCHOR_ENDPOINT` | `https://openrouter.ai/api/v1` | Any OpenAI-compatible endpoint |
-| `ANCHOR_MODEL` | `gpt-4o` | Model name |
-| `ANCHOR_MAX_TOKENS` | `4096` | Max response tokens |
-
 ## Slash Commands
 
 | Command | Description |
 |---|---|
-| `/setup` | Run interactive TUI to configure API key, model, endpoint |
+| `/setup` | Run interactive TUI to configure API key and model (also accessible via `anchor setup` subcommand) |
 | `/help` | Show available tools and commands |
 | `/exit` | Exit the application |
 | `/clear` | Clear the conversation history |
@@ -113,7 +103,7 @@ The resulting binary is ~12 MB, has no .NET runtime dependency, and starts insta
 ```
 AnchorCli/
 ├── Program.cs              # Entry point + REPL loop + AI client setup
-├── AnchorConfig.cs        # Environment variable configuration
+├── AnchorConfig.cs        # JSON file-based configuration (~APPDATA~\anchor\config.json)
 ├── ContextCompactor.cs    # Conversation history compression
 ├── AppJsonContext.cs      # Source-generated JSON context (AOT)
 ├── Hashline/
@@ -137,7 +127,7 @@ AnchorCli/
 
 ## How It Works
 
-1. **Setup**: Configure API credentials via environment variables or `/setup` command
+1. **Setup**: Configure API credentials via the `/setup` command (or `anchor setup` subcommand)
 2. **REPL Loop**: You interact with the AI through a conversational interface
 3. **Tool Calling**: The AI can call any of the available tools to read/edit files, manage directories, or execute commands
 4. **Hashline Validation**: All file edits are validated using the Hashline technique to ensure precision