feat(#589): wire WebToolsConfig to web search/fetch tools (#663)

* docs(config): update WebToolsConfig provider names

- Update doc comments to match actual implementation
- Document env var fallback behavior
- Change provider names from brave/searxng/google to exa/kimi_search

Part of: #589

* feat(tools): add from_config() to WebSearchTool

- Add from_config() method that accepts Option<&WebToolsConfig>
- Support provider selection via config.search_provider
- Fall back to environment variables when config not provided
- Extract from_env_inner() helper for code reuse
- Maintain backward compatibility with existing new() and from_env()

Part of: #589

* feat(tools): add from_config() to WebFetchTool

- Add from_config() method that accepts Option<&WebToolsConfig>
- Support fetch mode selection via config.fetch_mode
- Defaults to raw when config not provided or mode not specified
- Maintain backward compatibility with existing new() and with_mode()

Part of: #589

* feat(tools): wire config through to web tools

- Update create_default_registry signature to accept Option<&WebToolsConfig>
- Use from_config() instead of new() for web tools in registry
- Update main.rs to pass config.tools.web to registry creation
- Both agent mode and gateway mode now respect web_tools config

Part of: #589

* test(tools): add unit tests for from_config methods

- Add provider_name() test helper to WebSearchTool
- Add tests for WebSearchTool::from_config():
  - test_web_search_from_config_exa
  - test_web_search_from_config_kimi
  - test_web_search_from_config_fallback
  - test_web_search_from_config_unknown_provider
- Add tests for WebFetchTool::from_config():
  - test_web_fetch_from_config_raw
  - test_web_fetch_from_config_readability
  - test_web_fetch_from_config_fallback
  - test_web_fetch_from_config_none_mode

Part of: #589

* test(tools): add integration tests for config wiring

- Add test_web_tools_config_wired_to_registry
- Add test_registry_without_web_tools_config
- Add test_all_expected_tools_registered
- Verify config values flow through to tool registry

Part of: #589

Author: AlexMikhalev

crates/terraphim_tinyclaw/src/config.rs

@@ -414,10 +414,15 @@ fn default_shell_timeout() -> u64 {
 /// Web tools configuration.
 #[derive(Debug, Clone, Deserialize, Serialize)]
 pub struct WebToolsConfig {
-    /// Web search provider ("brave", "searxng", "google").
+    /// Web search provider ("exa", "kimi_search").
+    ///
+    /// If not specified, falls back to environment variables
+    /// (EXA_API_KEY or KIMI_API_KEY).
     pub search_provider: Option<String>,
 
-    /// Web fetch mode ("readability", "raw").
+    /// Web fetch mode ("raw", "readability").
+    ///
+    /// Defaults to "raw" if not specified.
     pub fetch_mode: Option<String>,
 }
 

crates/terraphim_tinyclaw/src/main.rs

@@ -169,7 +169,11 @@ async fn run_agent_mode(config: Config, system_prompt_path: Option<PathBuf>) ->
     let sessions = Arc::new(tokio::sync::Mutex::new(SessionManager::new(sessions_dir)));
 
     // Create tool registry with session manager
-    let tools = Arc::new(create_default_registry(Some(sessions.clone())));
+    let web_tools_config = config.tools.web.as_ref();
+    let tools = Arc::new(create_default_registry(
+        Some(sessions.clone()),
+        web_tools_config,
+    ));
 
     // Create hybrid LLM router
     let proxy_config = ProxyClientConfig {
@@ -222,7 +226,11 @@ async fn run_gateway_mode(config: Config) -> anyhow::Result<()> {
     let sessions = Arc::new(tokio::sync::Mutex::new(SessionManager::new(sessions_dir)));
 
     // Create tool registry with session manager
-    let tools = Arc::new(create_default_registry(Some(sessions.clone())));
+    let web_tools_config = config.tools.web.as_ref();
+    let tools = Arc::new(create_default_registry(
+        Some(sessions.clone()),
+        web_tools_config,
+    ));
 
     // Create hybrid LLM router
     let proxy_config = ProxyClientConfig {

crates/terraphim_tinyclaw/src/tools/mod.rs

@@ -149,8 +149,13 @@ impl Default for ToolRegistry {
 }
 
 /// Create a standard tool registry with all default tools.
+///
+/// # Arguments
+/// * `sessions` - Optional session manager for session-aware tools
+/// * `web_tools_config` - Optional web tools configuration
 pub fn create_default_registry(
     sessions: Option<std::sync::Arc<tokio::sync::Mutex<crate::session::SessionManager>>>,
+    web_tools_config: Option<&crate::config::WebToolsConfig>,
 ) -> ToolRegistry {
     use crate::tools::edit::EditTool;
     use crate::tools::filesystem::FilesystemTool;
@@ -163,8 +168,8 @@ pub fn create_default_registry(
     registry.register(Box::new(FilesystemTool::new()));
     registry.register(Box::new(EditTool::new()));
     registry.register(Box::new(ShellTool::new()));
-    registry.register(Box::new(WebSearchTool::new()));
-    registry.register(Box::new(WebFetchTool::new()));
+    registry.register(Box::new(WebSearchTool::from_config(web_tools_config)));
+    registry.register(Box::new(WebFetchTool::from_config(web_tools_config)));
     registry.register(Box::new(VoiceTranscribeTool::new()));
 
     // Register session tools if SessionManager is provided

crates/terraphim_tinyclaw/src/tools/web.rs

@@ -246,8 +246,47 @@ impl WebSearchTool {
         Self { provider }
     }
 
+    /// Get the provider name.
+    #[cfg(test)]
+    fn provider_name(&self) -> &str {
+        self.provider.name()
+    }
+
     /// Create from environment variables.
     pub fn from_env() -> Self {
+        Self::from_env_inner()
+    }
+
+    /// Create from configuration.
+    ///
+    /// If config specifies a search provider, uses it.
+    /// Otherwise falls back to environment variables (same as `new()`).
+    ///
+    /// # Arguments
+    /// * `config` - Optional web tools configuration
+    ///
+    /// # Supported Providers
+    /// - "exa" - Exa search API
+    /// - "kimi_search" - Kimi search API
+    pub fn from_config(config: Option<&crate::config::WebToolsConfig>) -> Self {
+        match config {
+            Some(cfg) => match cfg.search_provider.as_deref() {
+                Some("exa") => {
+                    let api_key = std::env::var("EXA_API_KEY").ok().filter(|k| !k.is_empty());
+                    Self::with_provider(Box::new(ExaProvider::new(api_key)))
+                }
+                Some("kimi_search") => {
+                    let api_key = std::env::var("KIMI_API_KEY").ok().filter(|k| !k.is_empty());
+                    Self::with_provider(Box::new(KimiSearchProvider::new(api_key)))
+                }
+                Some(_) | None => Self::from_env_inner(),
+            },
+            None => Self::from_env_inner(),
+        }
+    }
+
+    /// Internal helper to create from environment variables.
+    fn from_env_inner() -> Self {
         // Check for Exa API key
         if let Ok(api_key) = std::env::var("EXA_API_KEY") {
             if !api_key.is_empty() {
@@ -345,6 +384,28 @@ impl WebFetchTool {
         }
     }
 
+    /// Create from configuration.
+    ///
+    /// If config specifies a fetch mode, uses it.
+    /// Otherwise defaults to "raw".
+    ///
+    /// # Arguments
+    /// * `config` - Optional web tools configuration
+    ///
+    /// # Supported Modes
+    /// - "raw" - Fetch raw HTML
+    /// - "readability" - Extract readable content
+    pub fn from_config(config: Option<&crate::config::WebToolsConfig>) -> Self {
+        let mode = config
+            .and_then(|c| c.fetch_mode.clone())
+            .unwrap_or_else(|| "raw".to_string());
+
+        Self {
+            client: Client::new(),
+            mode,
+        }
+    }
+
     /// Fetch content from a URL.
     async fn fetch(&self, url: &str) -> Result<String, ToolError> {
         log::info!("Fetching URL: {} (mode: {})", url, self.mode);
@@ -492,4 +553,92 @@ mod tests {
         let provider = PlaceholderProvider;
         assert_eq!(provider.name(), "placeholder");
     }
+
+    #[test]
+    fn test_web_search_from_config_exa() {
+        let config = crate::config::WebToolsConfig {
+            search_provider: Some("exa".to_string()),
+            fetch_mode: None,
+        };
+
+        let tool = WebSearchTool::from_config(Some(&config));
+        // The provider name should reflect the configured provider
+        assert_eq!(tool.provider_name(), "exa");
+    }
+
+    #[test]
+    fn test_web_search_from_config_kimi() {
+        let config = crate::config::WebToolsConfig {
+            search_provider: Some("kimi_search".to_string()),
+            fetch_mode: None,
+        };
+
+        let tool = WebSearchTool::from_config(Some(&config));
+        assert_eq!(tool.provider_name(), "kimi_search");
+    }
+
+    #[test]
+    fn test_web_search_from_config_fallback() {
+        // When config is None, should fall back to env-based selection
+        // or placeholder if no env vars are set
+        let tool = WebSearchTool::from_config(None);
+        // Provider name should be one of the valid options
+        let name = tool.provider_name();
+        assert!(name == "exa" || name == "kimi_search" || name == "placeholder");
+    }
+
+    #[test]
+    fn test_web_search_from_config_unknown_provider() {
+        // Unknown provider should fall back to env-based selection
+        let config = crate::config::WebToolsConfig {
+            search_provider: Some("unknown_provider".to_string()),
+            fetch_mode: None,
+        };
+
+        let tool = WebSearchTool::from_config(Some(&config));
+        // Should fall back to env-based or placeholder
+        let name = tool.provider_name();
+        assert!(name == "exa" || name == "kimi_search" || name == "placeholder");
+    }
+
+    #[test]
+    fn test_web_fetch_from_config_raw() {
+        let config = crate::config::WebToolsConfig {
+            search_provider: None,
+            fetch_mode: Some("raw".to_string()),
+        };
+
+        let tool = WebFetchTool::from_config(Some(&config));
+        assert_eq!(tool.mode, "raw");
+    }
+
+    #[test]
+    fn test_web_fetch_from_config_readability() {
+        let config = crate::config::WebToolsConfig {
+            search_provider: None,
+            fetch_mode: Some("readability".to_string()),
+        };
+
+        let tool = WebFetchTool::from_config(Some(&config));
+        assert_eq!(tool.mode, "readability");
+    }
+
+    #[test]
+    fn test_web_fetch_from_config_fallback() {
+        // When config is None, should default to "raw"
+        let tool = WebFetchTool::from_config(None);
+        assert_eq!(tool.mode, "raw");
+    }
+
+    #[test]
+    fn test_web_fetch_from_config_none_mode() {
+        // When config has None for fetch_mode, should default to "raw"
+        let config = crate::config::WebToolsConfig {
+            search_provider: Some("exa".to_string()),
+            fetch_mode: None,
+        };
+
+        let tool = WebFetchTool::from_config(Some(&config));
+        assert_eq!(tool.mode, "raw");
+    }
 }

crates/terraphim_tinyclaw/tests/config_wiring.rs

@@ -0,0 +1,78 @@
+//! Integration tests for config wiring to tools.
+//!
+//! Tests that configuration values from files are properly passed to tools.
+
+use terraphim_tinyclaw::config::{Config, ToolsConfig, WebToolsConfig};
+use terraphim_tinyclaw::tools::create_default_registry;
+
+/// Test that web tools configuration is wired through to the registry.
+#[test]
+fn test_web_tools_config_wired_to_registry() {
+    // Create a config with specific web tools settings
+    let config = Config {
+        tools: ToolsConfig {
+            web: Some(WebToolsConfig {
+                search_provider: Some("exa".to_string()),
+                fetch_mode: Some("readability".to_string()),
+            }),
+            ..Default::default()
+        },
+        ..Default::default()
+    };
+
+    // Create registry with the web tools config
+    let web_tools_config = config.tools.web.as_ref();
+    let registry = create_default_registry(None, web_tools_config);
+
+    // Verify web_search tool is present
+    let web_search = registry.get("web_search");
+    assert!(web_search.is_some(), "web_search tool should be registered");
+
+    // Verify web_fetch tool is present
+    let web_fetch = registry.get("web_fetch");
+    assert!(web_fetch.is_some(), "web_fetch tool should be registered");
+}
+
+/// Test that registry works with no web tools config.
+#[test]
+fn test_registry_without_web_tools_config() {
+    // Create registry without web tools config
+    let registry = create_default_registry(None, None);
+
+    // Verify web_search tool is still present (with defaults)
+    let web_search = registry.get("web_search");
+    assert!(
+        web_search.is_some(),
+        "web_search tool should be registered even without config"
+    );
+
+    // Verify web_fetch tool is still present (with defaults)
+    let web_fetch = registry.get("web_fetch");
+    assert!(
+        web_fetch.is_some(),
+        "web_fetch tool should be registered even without config"
+    );
+}
+
+/// Test that all expected tools are registered.
+#[test]
+fn test_all_expected_tools_registered() {
+    let registry = create_default_registry(None, None);
+
+    let expected_tools = [
+        "filesystem",
+        "edit",
+        "shell",
+        "web_search",
+        "web_fetch",
+        "voice_transcribe",
+    ];
+
+    for tool_name in &expected_tools {
+        assert!(
+            registry.get(tool_name).is_some(),
+            "Tool '{}' should be registered",
+            tool_name
+        );
+    }
+}