feat(phase10): complete MCP-only architecture migration

This commit completes Phase 10 of the MCP migration by removing all
direct provider usage from CLI/TUI and enforcing MCP-first architecture.

## Changes

### Core Architecture
- **main.rs**: Replaced OllamaProvider with RemoteMcpClient
  - Uses MCP server configuration from config.toml if available
  - Falls back to auto-discovery of MCP LLM server binary
- **agent_main.rs**: Unified provider and MCP client to single RemoteMcpClient
  - Simplifies initialization with Arc::clone pattern
  - All LLM communication now goes through MCP protocol

### Dependencies
- **Cargo.toml**: Removed owlen-ollama dependency from owlen-cli
  - CLI no longer knows about Ollama implementation details
  - Clean separation: only MCP servers use provider crates internally

### Tests
- **agent_tests.rs**: Updated all tests to use RemoteMcpClient
  - Replaced OllamaProvider::new() with RemoteMcpClient::new()
  - Updated test documentation to reflect MCP requirements
  - All tests compile and run successfully

### Examples
- **Removed**: custom_provider.rs, basic_chat.rs (deprecated)
- **Added**: mcp_chat.rs - demonstrates recommended MCP-based usage
  - Shows how to use RemoteMcpClient for LLM interactions
  - Includes model listing and chat request examples

### Cleanup
- Removed outdated TODO about MCP integration (now complete)
- Updated comments to reflect current MCP architecture

## Architecture

```
CLI/TUI → RemoteMcpClient (impl Provider)
          ↓ MCP Protocol (STDIO/HTTP/WS)
          MCP LLM Server → OllamaProvider → Ollama
```

## Benefits
- ✅ Clean separation of concerns
- ✅ CLI is protocol-agnostic (only knows MCP)
- ✅ Easier to add new LLM backends (just implement MCP server)
- ✅ All tests passing
- ✅ Full workspace builds successfully

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

crates/owlen-cli/Cargo.toml

@@ -24,7 +24,6 @@ required-features = ["chat-client"]
 
 [dependencies]
 owlen-core = { path = "../owlen-core" }
-owlen-ollama = { path = "../owlen-ollama" }
 # Optional TUI dependency, enabled by the "chat-client" feature.
 owlen-tui = { path = "../owlen-tui", optional = true }
 

crates/owlen-cli/src/agent_main.rs

@@ -11,11 +11,15 @@ use std::sync::Arc;
 use clap::Parser;
 use owlen_cli::agent::{AgentConfig, AgentExecutor};
 use owlen_core::mcp::remote_client::RemoteMcpClient;
-use owlen_ollama::OllamaProvider;
 
 /// Command‑line arguments for the agent binary.
 #[derive(Parser, Debug)]
-#[command(name = "owlen-agent", author, version, about = "Run the ReAct agent")]
+#[command(
+    name = "owlen-agent",
+    author,
+    version,
+    about = "Run the ReAct agent via MCP"
+)]
 struct Args {
     /// The initial user query.
     prompt: String,
@@ -31,11 +35,13 @@ struct Args {
 async fn main() -> anyhow::Result<()> {
     let args = Args::parse();
 
-    // Initialise the LLM provider (Ollama) – uses default local URL.
-    let provider = Arc::new(OllamaProvider::new("http://localhost:11434")?);
-    // Initialise the MCP client (remote LLM server) – this client also knows how
-    // to call the built‑in resource tools.
-    let mcp_client = Arc::new(RemoteMcpClient::new()?);
+    // Initialise the MCP LLM client – it implements Provider and talks to the
+    // MCP LLM server which wraps Ollama. This ensures all communication goes
+    // through the MCP architecture (Phase 10 requirement).
+    let provider = Arc::new(RemoteMcpClient::new()?);
+
+    // The MCP client also serves as the tool client for resource operations
+    let mcp_client = Arc::clone(&provider) as Arc<RemoteMcpClient>;
 
     let config = AgentConfig {
         max_iterations: args.max_iter,

crates/owlen-cli/src/main.rs

@@ -2,8 +2,10 @@
 
 use anyhow::Result;
 use clap::Parser;
-use owlen_core::{mode::Mode, session::SessionController, storage::StorageManager};
-use owlen_ollama::OllamaProvider;
+use owlen_core::{
+    mcp::remote_client::RemoteMcpClient, mode::Mode, session::SessionController,
+    storage::StorageManager, Provider,
+};
 use owlen_tui::tui_controller::{TuiController, TuiRequest};
 use owlen_tui::{config, ui, AppState, ChatApp, Event, EventHandler, SessionEvent};
 use std::io;
@@ -21,7 +23,7 @@ use ratatui::{prelude::CrosstermBackend, Terminal};
 /// Owlen - Terminal UI for LLM chat
 #[derive(Parser, Debug)]
 #[command(name = "owlen")]
-#[command(about = "Terminal UI for LLM chat with Ollama", long_about = None)]
+#[command(about = "Terminal UI for LLM chat via MCP", long_about = None)]
 struct Args {
     /// Start in code mode (enables all tools)
     #[arg(long, short = 'c')]
@@ -44,21 +46,16 @@ async fn main() -> Result<()> {
     let mut cfg = config::try_load_config().unwrap_or_default();
     // Disable encryption for CLI to avoid password prompts in this environment.
     cfg.privacy.encrypt_local_data = false;
-    // Determine provider configuration
-    let provider_name = cfg.general.default_provider.clone();
-    let provider_cfg = config::ensure_provider_config(&mut cfg, &provider_name).clone();
-    let provider_type = provider_cfg.provider_type.to_ascii_lowercase();
-    if provider_type != "ollama" && provider_type != "ollama-cloud" {
-        anyhow::bail!(
-            "Unsupported provider type '{}' configured for provider '{}'",
-            provider_cfg.provider_type,
-            provider_name,
-        );
-    }
-    let provider = Arc::new(OllamaProvider::from_config(
-        &provider_cfg,
-        Some(&cfg.general),
-    )?);
+
+    // Create MCP LLM client as the provider (replaces direct OllamaProvider usage)
+    let provider: Arc<dyn Provider> = if let Some(mcp_server) = cfg.mcp_servers.first() {
+        // Use configured MCP server if available
+        Arc::new(RemoteMcpClient::new_with_config(mcp_server)?)
+    } else {
+        // Fall back to default MCP LLM server discovery
+        Arc::new(RemoteMcpClient::new()?)
+    };
+
     let storage = Arc::new(StorageManager::new().await?);
     let controller =
         SessionController::new(provider, cfg, storage.clone(), tui_controller, false).await?;

crates/owlen-cli/tests/agent_tests.rs

@@ -9,7 +9,6 @@
 
 use owlen_cli::agent::{AgentConfig, AgentExecutor, LlmResponse};
 use owlen_core::mcp::remote_client::RemoteMcpClient;
-use owlen_ollama::OllamaProvider;
 use std::sync::Arc;
 
 #[tokio::test]
@@ -72,11 +71,11 @@ async fn test_react_parsing_with_multiline_thought() {
 }
 
 #[tokio::test]
-#[ignore] // Requires Ollama to be running
+#[ignore] // Requires MCP LLM server to be running
 async fn test_agent_single_tool_scenario() {
-    // This test requires a running Ollama instance and MCP server
-    let provider = Arc::new(OllamaProvider::new("http://localhost:11434").unwrap());
-    let mcp_client = Arc::new(RemoteMcpClient::new().unwrap());
+    // This test requires a running MCP LLM server (which wraps Ollama)
+    let provider = Arc::new(RemoteMcpClient::new().unwrap());
+    let mcp_client = Arc::clone(&provider) as Arc<RemoteMcpClient>;
 
     let config = AgentConfig {
         max_iterations: 5,
@@ -109,8 +108,8 @@ async fn test_agent_single_tool_scenario() {
 #[ignore] // Requires Ollama to be running
 async fn test_agent_multi_step_workflow() {
     // Test a query that requires multiple tool calls
-    let provider = Arc::new(OllamaProvider::new("http://localhost:11434").unwrap());
-    let mcp_client = Arc::new(RemoteMcpClient::new().unwrap());
+    let provider = Arc::new(RemoteMcpClient::new().unwrap());
+    let mcp_client = Arc::clone(&provider) as Arc<RemoteMcpClient>;
 
     let config = AgentConfig {
         max_iterations: 10,
@@ -141,8 +140,8 @@ async fn test_agent_multi_step_workflow() {
 #[tokio::test]
 #[ignore] // Requires Ollama
 async fn test_agent_iteration_limit() {
-    let provider = Arc::new(OllamaProvider::new("http://localhost:11434").unwrap());
-    let mcp_client = Arc::new(RemoteMcpClient::new().unwrap());
+    let provider = Arc::new(RemoteMcpClient::new().unwrap());
+    let mcp_client = Arc::clone(&provider) as Arc<RemoteMcpClient>;
 
     let config = AgentConfig {
         max_iterations: 2, // Very low limit to test enforcement
@@ -183,8 +182,8 @@ async fn test_agent_iteration_limit() {
 #[tokio::test]
 #[ignore] // Requires Ollama
 async fn test_agent_tool_budget_enforcement() {
-    let provider = Arc::new(OllamaProvider::new("http://localhost:11434").unwrap());
-    let mcp_client = Arc::new(RemoteMcpClient::new().unwrap());
+    let provider = Arc::new(RemoteMcpClient::new().unwrap());
+    let mcp_client = Arc::clone(&provider) as Arc<RemoteMcpClient>;
 
     let config = AgentConfig {
         max_iterations: 20,
@@ -224,12 +223,9 @@ async fn test_agent_tool_budget_enforcement() {
 // Helper function to create a test executor
 // For parsing tests, we don't need a real connection
 fn create_test_executor() -> AgentExecutor {
-    // Create dummy instances - the parse_response method doesn't actually use them
-    let provider = Arc::new(OllamaProvider::new("http://localhost:11434").unwrap());
-
     // For parsing tests, we can accept the error from RemoteMcpClient::new()
     // since we're only testing parse_response which doesn't use the MCP client
-    let mcp_client = match RemoteMcpClient::new() {
+    let provider = match RemoteMcpClient::new() {
         Ok(client) => Arc::new(client),
         Err(_) => {
             // If MCP server binary doesn't exist, parsing tests can still run
@@ -239,6 +235,8 @@ fn create_test_executor() -> AgentExecutor {
         }
     };
 
+    let mcp_client = Arc::clone(&provider) as Arc<RemoteMcpClient>;
+
     let config = AgentConfig::default();
     AgentExecutor::new(provider, mcp_client, config, None)
 }

crates/owlen-tui/src/chat_app.rs

@@ -311,7 +311,7 @@ impl ChatApp {
     pub async fn set_mode(&mut self, mode: owlen_core::mode::Mode) {
         self.operating_mode = mode;
         self.status = format!("Switched to {} mode", mode);
-        // TODO: Update MCP client mode when MCP integration is fully implemented
+        // Mode switching is handled by the SessionController's tool filtering
     }
 
     pub(crate) fn model_selector_items(&self) -> &[ModelSelectorItem] {

examples/basic_chat.rs

@@ -1,30 +0,0 @@
-// This example demonstrates a basic chat interaction without the TUI.
-
-use owlen_core::model::Model;
-use owlen_core::provider::Provider;
-use owlen_core::session::Session;
-use owlen_ollama::OllamaProvider; // Assuming you have an Ollama provider
-
-#[tokio::main]
-async fn main() -> Result<(), anyhow::Error> {
-    // This example requires a running Ollama instance.
-    // Make sure you have a model available, e.g., `ollama pull llama2`
-
-    let provider = OllamaProvider;
-    let model = Model::new("llama2"); // Change to a model you have
-    let mut session = Session::new("basic-chat-session");
-
-    println!("Starting basic chat with model: {}", model.name);
-
-    let user_message = "What is the capital of France?";
-    session.add_message("user", user_message);
-    println!("User: {}", user_message);
-
-    // Send the chat to the provider
-    let response = provider.chat(&session, &model).await?;
-
-    session.add_message("bot", &response);
-    println!("Bot: {}", response);
-
-    Ok(())
-}

examples/custom_provider.rs

@@ -1,45 +0,0 @@
-// This example demonstrates how to implement a custom provider.
-
-use async_trait::async_trait;
-use owlen_core::model::Model;
-use owlen_core::provider::Provider;
-use owlen_core::session::Session;
-
-// Define a struct for your custom provider.
-pub struct MyCustomProvider;
-
-// Implement the `Provider` trait for your struct.
-#[async_trait]
-impl Provider for MyCustomProvider {
-    fn name(&self) -> &str {
-        "custom-provider"
-    }
-
-    async fn chat(&self, session: &Session, model: &Model) -> Result<String, anyhow::Error> {
-        println!(
-            "Custom provider received chat request for model: {}",
-            model.name
-        );
-        // In a real implementation, you would send the session data to an API.
-        let message_count = session.get_messages().len();
-        Ok(format!(
-            "This is a custom response. You have {} messages in your session.",
-            message_count
-        ))
-    }
-}
-
-#[tokio::main]
-async fn main() -> Result<(), anyhow::Error> {
-    let provider = MyCustomProvider;
-    let model = Model::new("custom-model");
-    let mut session = Session::new("custom-session");
-
-    session.add_message("user", "Hello, custom provider!");
-
-    let response = provider.chat(&session, &model).await?;
-
-    println!("Provider response: {}", response);
-
-    Ok(())
-}

examples/mcp_chat.rs

@@ -0,0 +1,71 @@
+//! Example demonstrating MCP-based chat interaction.
+//!
+//! This example shows the recommended way to interact with LLMs via the MCP architecture.
+//! It uses `RemoteMcpClient` which communicates with the MCP LLM server.
+//!
+//! Prerequisites:
+//! - Build the MCP LLM server: `cargo build --release -p owlen-mcp-llm-server`
+//! - Ensure Ollama is running with a model available
+
+use owlen_core::{
+    mcp::remote_client::RemoteMcpClient,
+    types::{ChatParameters, ChatRequest, Message, Role},
+    Provider,
+};
+use std::sync::Arc;
+
+#[tokio::main]
+async fn main() -> Result<(), anyhow::Error> {
+    println!("🦉 Owlen MCP Chat Example\n");
+
+    // Create MCP client - this will spawn/connect to the MCP LLM server
+    println!("Connecting to MCP LLM server...");
+    let client = Arc::new(RemoteMcpClient::new()?);
+    println!("✓ Connected\n");
+
+    // List available models
+    println!("Fetching available models...");
+    let models = client.list_models().await?;
+    println!("Available models:");
+    for model in &models {
+        println!("  - {} ({})", model.name, model.provider);
+    }
+    println!();
+
+    // Select first available model or default
+    let model_name = models
+        .first()
+        .map(|m| m.id.clone())
+        .unwrap_or_else(|| "llama3.2:latest".to_string());
+    println!("Using model: {}\n", model_name);
+
+    // Create a simple chat request
+    let user_message = "What is the capital of France? Please be concise.";
+    println!("User: {}", user_message);
+
+    let request = ChatRequest {
+        model: model_name,
+        messages: vec![Message::new(Role::User, user_message.to_string())],
+        parameters: ChatParameters {
+            temperature: Some(0.7),
+            max_tokens: Some(100),
+            stream: false,
+            extra: std::collections::HashMap::new(),
+        },
+        tools: None,
+    };
+
+    // Send request and get response
+    println!("\nAssistant: ");
+    let response = client.chat(request).await?;
+    println!("{}", response.message.content);
+
+    if let Some(usage) = response.usage {
+        println!(
+            "\n📊 Tokens: {} prompt + {} completion = {} total",
+            usage.prompt_tokens, usage.completion_tokens, usage.total_tokens
+        );
+    }
+
+    Ok(())
+}