feat(commit-ai): enhance commit message generation and caching

Improve the commit message generation process by adding a caching mechanism and refining the input handling. This change aims to enhance performance and prevent redundant computations when generating commit messages based on staged changes.

- **src-tauri/crates/commit_ai/src/cloud.rs**:
  - Introduced `looks_like_diff_echo` function to detect if the model's output is a diff instead of a commit message.
  - Updated `openai_compatible_request` and `generate_anthropic` to utilize the new function for error handling.

- **src-tauri/crates/commit_ai/src/lib.rs**:
  - Added `LocalGenerationProfile` enum for managing different generation profiles.
  - Implemented caching for generated messages to avoid redundant processing.
  - Updated `generate_commit_message` to incorporate caching logic.

- **src-tauri/src/git.rs**:
  - Added `staged_diff_local` function to handle local profile generation and exclude specific lock files from the diff.
  - Modified `commit_ai_generate` to accept and process the local generation profile.

- **src/App.svelte**:
  - Added `lastLocalAiGeneratedMessage` state to track the last generated message and prevent unnecessary updates.

- **.claude/settings.local.json**:
  - Updated settings to include additional commands for better functionality.

src-tauri/crates/commit_ai/src/cloud.rs

@@ -2,7 +2,7 @@ use std::time::Duration;
 
 use serde::{Deserialize, Serialize};
 
-use crate::{build_messages, sanitize_message};
+use crate::{build_messages, looks_like_diff_echo, sanitize_message};
 
 // Generous sizing so a detailed body with bullet points isn't cut off.
 const DEFAULT_MAX_TOKENS: u32 = 1500;
@@ -56,8 +56,14 @@ async fn openai_compatible_request(
     let body = OpenAiRequest {
         model: model.to_string(),
         messages: vec![
-            OpenAiMessage { role: "system", content: system },
-            OpenAiMessage { role: "user", content: user },
+            OpenAiMessage {
+                role: "system",
+                content: system,
+            },
+            OpenAiMessage {
+                role: "user",
+                content: user,
+            },
         ],
         temperature: 0.3,
     };
@@ -82,17 +88,22 @@ async fn openai_compatible_request(
         return Err(format!("API error ({status}): {text}"));
     }
 
-    let parsed: OpenAiResponse = serde_json::from_str(&text)
-        .map_err(|err| format!("Could not process response: {err}"))?;
+    let parsed: OpenAiResponse =
+        serde_json::from_str(&text).map_err(|err| format!("Could not process response: {err}"))?;
 
-    parsed
+    let message = parsed
         .choices
         .into_iter()
         .next()
         .and_then(|choice| choice.message.content)
         .map(|content| sanitize_message(&content))
         .filter(|content| !content.is_empty())
-        .ok_or_else(|| "The model did not return a response.".to_string())
+        .ok_or_else(|| "The model did not return a response.".to_string())?;
+
+    if looks_like_diff_echo(&message) {
+        return Err("The model returned the diff instead of a commit message.".to_string());
+    }
+    Ok(message)
 }
 
 pub async fn generate_openai(
@@ -168,7 +179,10 @@ pub async fn generate_anthropic(
         model: model.to_string(),
         max_tokens: DEFAULT_MAX_TOKENS,
         system,
-        messages: vec![AnthropicMessage { role: "user", content: user }],
+        messages: vec![AnthropicMessage {
+            role: "user",
+            content: user,
+        }],
     };
 
     let client = http_client()?;
@@ -190,14 +204,19 @@ pub async fn generate_anthropic(
         return Err(format!("API error ({status}): {text}"));
     }
 
-    let parsed: AnthropicResponse = serde_json::from_str(&text)
-        .map_err(|err| format!("Could not process response: {err}"))?;
+    let parsed: AnthropicResponse =
+        serde_json::from_str(&text).map_err(|err| format!("Could not process response: {err}"))?;
 
-    parsed
+    let message = parsed
         .content
         .into_iter()
         .find_map(|block| block.text)
         .map(|text| sanitize_message(&text))
         .filter(|text| !text.is_empty())
-        .ok_or_else(|| "The model did not return a response.".to_string())
+        .ok_or_else(|| "The model did not return a response.".to_string())?;
+
+    if looks_like_diff_echo(&message) {
+        return Err("The model returned the diff instead of a commit message.".to_string());
+    }
+    Ok(message)
 }