refactor(commit-ai): remove local models and simplify AI flow

Remove local on-device model support and related IPC commands,
consolidating commit-generation to cloud providers and simplifying the
AI crate surface. Local-specific types, generation profiles, caching,
and the local prompt builder were removed while message sanitization and
diff-echo detection were preserved. Also harden repository handling and
runtime: unborn HEADs are handled gracefully so empty repos still report
files, Git LFS sync is skipped for repositories without commits, and
tokio runtime features were enabled.

- Remove local model engine, load/status commands, and local profile code
- Handle unborn HEAD and skip LFS sync for repos without commits
- Enable tokio runtime features and route AI generation to cloud only

src-tauri/Cargo.toml

@@ -20,7 +20,7 @@ tauri-plugin-dialog = "=2.7.0"
 tauri-plugin-aptabase = "1.0"
 keyring = { version = "3", features = ["apple-native", "windows-native", "async-secret-service", "crypto-rust", "async-io"] }
 commit_ai = { path = "crates/commit_ai" }
-tokio = "1.52.3"
+tokio = { version = "1.52.3", features = ["macros", "rt-multi-thread"] }
 log = "0.4"
 reqwest = { version = "0.12", default-features = false, features = ["blocking", "json", "rustls-tls"] }
 sysinfo = { version = "=0.38.3", default-features = false, features = ["system"] }

src-tauri/crates/commit_ai/Cargo.toml

@@ -5,8 +5,6 @@ version = "0.1.0"
 edition = "2024"
 
 [dependencies]
-mistralrs = "0.8"
-tokio = { version = "1", features = ["sync"] }
 serde = { version = "1", features = ["derive"] }
 serde_json = "1"
 reqwest = { version = "0.12", default-features = false, features = ["json", "rustls-tls"] }

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

@@ -5,318 +5,8 @@ pub use cloud::{
     review_openai, split_anthropic, split_custom, split_openai,
 };
 
-use std::{
+/// Strip a wrapping code fence and wrapping quotes so the result can go straight into
-    collections::hash_map::DefaultHasher,
+/// the commit-message box.
-    hash::{Hash, Hasher},
-    sync::Arc,
-};
-
-use mistralrs::{GgufModelBuilder, Model, RequestBuilder, TextMessageRole};
-use tokio::sync::RwLock;
-
-/// One selectable local (on-device) model. Larger models produce better commit messages
-/// but take longer to download (first run only, then cached) and run slower on CPU.
-#[derive(Debug, Clone, serde::Serialize)]
-pub struct LocalModelOption {
-    pub id: &'static str,
-    pub label: &'static str,
-    pub approx_size_mb: u32,
-    repo: &'static str,
-    file: &'static str,
-    tokenizer_repo: &'static str,
-}
-
-pub const DEFAULT_LOCAL_MODEL_ID: &str = "qwen2.5-0.5b";
-
-pub const LOCAL_MODELS: &[LocalModelOption] = &[
-    LocalModelOption {
-        id: "qwen2.5-0.5b",
-        label: "Qwen2.5 0.5B Instruct — fast, lower quality",
-        approx_size_mb: 490,
-        repo: "Qwen/Qwen2.5-0.5B-Instruct-GGUF",
-        file: "qwen2.5-0.5b-instruct-q4_k_m.gguf",
-        tokenizer_repo: "Qwen/Qwen2.5-0.5B-Instruct",
-    },
-    LocalModelOption {
-        id: "qwen2.5-1.5b",
-        label: "Qwen2.5 1.5B Instruct — recommended",
-        approx_size_mb: 1050,
-        repo: "Qwen/Qwen2.5-1.5B-Instruct-GGUF",
-        file: "qwen2.5-1.5b-instruct-q4_k_m.gguf",
-        tokenizer_repo: "Qwen/Qwen2.5-1.5B-Instruct",
-    },
-    LocalModelOption {
-        id: "qwen2.5-3b",
-        label: "Qwen2.5 3B Instruct — best quality, slower",
-        approx_size_mb: 2100,
-        repo: "Qwen/Qwen2.5-3B-Instruct-GGUF",
-        file: "qwen2.5-3b-instruct-q4_k_m.gguf",
-        tokenizer_repo: "Qwen/Qwen2.5-3B-Instruct",
-    },
-];
-
-fn find_local_model(model_id: &str) -> Option<&'static LocalModelOption> {
-    LOCAL_MODELS.iter().find(|option| option.id == model_id)
-}
-
-#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, serde::Serialize, serde::Deserialize)]
-#[serde(rename_all = "lowercase")]
-pub enum LocalGenerationProfile {
-    Fast,
-    Balanced,
-    Detailed,
-}
-
-impl Default for LocalGenerationProfile {
-    fn default() -> Self {
-        Self::Fast
-    }
-}
-
-impl LocalGenerationProfile {
-    pub fn from_id(value: Option<&str>) -> Self {
-        match value
-            .unwrap_or_default()
-            .trim()
-            .to_ascii_lowercase()
-            .as_str()
-        {
-            "balanced" => Self::Balanced,
-            "detailed" => Self::Detailed,
-            _ => Self::Fast,
-        }
-    }
-
-    pub fn diff_unified_context(self) -> &'static str {
-        match self {
-            Self::Fast => "--unified=1",
-            Self::Balanced => "--unified=2",
-            Self::Detailed => "--unified=3",
-        }
-    }
-
-    fn max_diff_chars(self) -> usize {
-        match self {
-            Self::Fast => 8_000,
-            Self::Balanced => 12_000,
-            Self::Detailed => 24_000,
-        }
-    }
-
-    fn max_output_tokens(self) -> usize {
-        match self {
-            Self::Fast => 160,
-            Self::Balanced => 360,
-            Self::Detailed => 750,
-        }
-    }
-}
-
-#[derive(Debug, Clone, Copy, PartialEq, Eq, serde::Serialize)]
-#[serde(rename_all = "lowercase")]
-pub enum CommitAiPhase {
-    /// Nothing has been requested yet.
-    Idle,
-    /// Downloading (first run only, then cached by hf-hub) and/or loading into memory.
-    Loading,
-    Ready,
-    Error,
-}
-
-#[derive(Debug, Clone, serde::Serialize)]
-pub struct CommitAiStatus {
-    pub phase: CommitAiPhase,
-    pub model_id: Option<String>,
-    pub error: Option<String>,
-}
-
-struct Inner {
-    phase: CommitAiPhase,
-    model_id: Option<String>,
-    error: Option<String>,
-    model: Option<Arc<Model>>,
-    cache: Option<GenerationCache>,
-}
-
-#[derive(Debug, Clone, PartialEq, Eq)]
-struct GenerationCacheKey {
-    model_id: String,
-    profile: LocalGenerationProfile,
-    input_hash: u64,
-}
-
-#[derive(Debug, Clone)]
-struct GenerationCache {
-    key: GenerationCacheKey,
-    message: String,
-}
-
-/// Manages the local (on-device) model only. Cloud providers are stateless HTTP calls
-/// (see [`cloud`]) and don't need this — there's nothing to download or keep loaded.
-#[derive(Clone)]
-pub struct CommitAiEngine {
-    inner: Arc<RwLock<Inner>>,
-}
-
-impl Default for CommitAiEngine {
-    fn default() -> Self {
-        Self {
-            inner: Arc::new(RwLock::new(Inner {
-                phase: CommitAiPhase::Idle,
-                model_id: None,
-                error: None,
-                model: None,
-                cache: None,
-            })),
-        }
-    }
-}
-
-impl CommitAiEngine {
-    pub fn new() -> Self {
-        Self::default()
-    }
-
-    pub async fn status(&self) -> CommitAiStatus {
-        let guard = self.inner.read().await;
-        CommitAiStatus {
-            phase: guard.phase,
-            model_id: guard.model_id.clone(),
-            error: guard.error.clone(),
-        }
-    }
-
-    /// Downloads (first run only; hf-hub caches the files afterwards) and loads the given
-    /// local model. Safe to call repeatedly — a call for the model that's already
-    /// ready/loading is a no-op; a call for a *different* model switches to it (the
-    /// previous one is dropped once no generation is still using it).
-    pub async fn ensure_loaded(&self, model_id: &str) {
-        {
-            let guard = self.inner.read().await;
-            let same_model = guard.model_id.as_deref() == Some(model_id);
-            if same_model && matches!(guard.phase, CommitAiPhase::Ready | CommitAiPhase::Loading) {
-                return;
-            }
-        }
-
-        let Some(option) = find_local_model(model_id) else {
-            let mut guard = self.inner.write().await;
-            guard.phase = CommitAiPhase::Error;
-            guard.model_id = Some(model_id.to_string());
-            guard.error = Some(format!("Unknown local model: {model_id}"));
-            guard.cache = None;
-            return;
-        };
-
-        {
-            let mut guard = self.inner.write().await;
-            guard.phase = CommitAiPhase::Loading;
-            guard.model_id = Some(model_id.to_string());
-            guard.error = None;
-            guard.model = None;
-            guard.cache = None;
-        }
-
-        let result = GgufModelBuilder::new(option.repo, vec![option.file])
-            .with_tok_model_id(option.tokenizer_repo)
-            .with_logging()
-            .build()
-            .await;
-
-        let mut guard = self.inner.write().await;
-        // If the user switched to yet another model while this one was loading, drop this
-        // (now stale) result instead of overwriting the newer request's state.
-        if guard.model_id.as_deref() != Some(model_id) {
-            return;
-        }
-        match result {
-            Ok(model) => {
-                guard.model = Some(Arc::new(model));
-                guard.phase = CommitAiPhase::Ready;
-                guard.error = None;
-                guard.cache = None;
-            }
-            Err(err) => {
-                guard.phase = CommitAiPhase::Error;
-                guard.error = Some(err.to_string());
-                guard.cache = None;
-            }
-        }
-    }
-
-    pub async fn generate_commit_message(
-        &self,
-        diff: &str,
-        notes: Option<&str>,
-        profile: LocalGenerationProfile,
-    ) -> Result<String, String> {
-        let (model, cache_key) = {
-            let guard = self.inner.read().await;
-            match (guard.phase, &guard.model) {
-                (CommitAiPhase::Ready, Some(model)) => {
-                    let cache_key = GenerationCacheKey {
-                        model_id: guard.model_id.clone().unwrap_or_default(),
-                        profile,
-                        input_hash: generation_input_hash(diff, notes),
-                    };
-                    if let Some(cache) = &guard.cache {
-                        if cache.key == cache_key {
-                            return Ok(cache.message.clone());
-                        }
-                    }
-                    (model.clone(), cache_key)
-                }
-                _ => return Err("The local AI model is not ready yet.".to_string()),
-            }
-        };
-
-        let (system, user) = build_local_messages(diff, notes, profile)?;
-        let request = RequestBuilder::new()
-            .set_sampler_max_len(profile.max_output_tokens())
-            .add_message(TextMessageRole::System, system)
-            .add_message(TextMessageRole::User, user);
-
-        let response = model
-            .send_chat_request(request)
-            .await
-            .map_err(|err| err.to_string())?;
-
-        let content = response
-            .choices
-            .first()
-            .and_then(|choice| choice.message.content.clone())
-            .ok_or_else(|| "The model did not return a response.".to_string())?;
-
-        let message = sanitize_message(&content);
-        if message.is_empty() {
-            return Err("The model did not return a response.".to_string());
-        }
-        if looks_like_diff_echo(&message) {
-            return Err(
-                "The local model returned the diff instead of a commit message. Try a larger local model (1.5B or 3B) or a cloud provider.".to_string(),
-            );
-        }
-        {
-            let mut guard = self.inner.write().await;
-            guard.cache = Some(GenerationCache {
-                key: cache_key,
-                message: message.clone(),
-            });
-        }
-        Ok(message)
-    }
-}
-
-fn generation_input_hash(diff: &str, notes: Option<&str>) -> u64 {
-    let mut hasher = DefaultHasher::new();
-    diff.hash(&mut hasher);
-    notes.unwrap_or("").hash(&mut hasher);
-    hasher.finish()
-}
-
-/// Models occasionally ignore the "no code fences" instruction (small local models
-/// especially) — strip a wrapping ``` fence and wrapping quotes so the result can go
-/// straight into the commit-message box.
 pub(crate) fn sanitize_message(raw: &str) -> String {
     let mut text = raw.trim().to_string();
     if text.starts_with("```") {
@@ -333,9 +23,9 @@ pub(crate) fn sanitize_message(raw: &str) -> String {
     trimmed.to_string()
 }
 
-/// Weak models (small local ones especially) sometimes just echo the prompt's diff
+/// Some models echo the prompt's diff sections instead of writing a commit message.
-/// sections back instead of writing a commit message. Catch that so the UI can show a
+/// Catch that so the UI can show a clear error instead of dumping raw diff text into
-/// clear error instead of dumping raw diff text into the commit-message box.
+/// the commit-message box.
 pub(crate) fn looks_like_diff_echo(message: &str) -> bool {
     let lower = message.to_ascii_lowercase();
     lower.contains("diff --git")
@@ -358,55 +48,12 @@ fn truncate_at_char_boundary(input: &str, max_chars: usize) -> String {
     format!("{}\n\n[... diff truncated ...]", &input[..cut])
 }
 
-pub(crate) fn build_local_messages(
-    diff: &str,
-    notes: Option<&str>,
-    profile: LocalGenerationProfile,
-) -> Result<(String, String), String> {
-    if diff.trim().is_empty() {
-        return Err("No staged changes available for a commit message.".to_string());
-    }
-
-    let diff = truncate_at_char_boundary(diff, profile.max_diff_chars());
-    // Appended to every profile below: small local models occasionally just echo the input
-    // (the "Staged files:" / "Diff stat:" / "Detailed diff:" sections built in git.rs's
-    // `staged_diff_local`) instead of writing a new commit message. Naming those exact
-    // section headers here makes the failure mode explicit enough for weak models to avoid.
-    const ANTI_ECHO: &str = " Never repeat, quote, or paraphrase the diff or its headers — \
-do not include 'diff --git', '@@', 'Staged files:', 'Diff stat:', or 'Detailed diff:' \
-anywhere in your answer.";
-    let system = match profile {
-        LocalGenerationProfile::Fast => {
-            format!(
-                "You generate Git commit messages. Respond only with one Conventional Commits subject line: <type>(<scope>): <subject>. Max 72 characters. No body, bullets, preamble, explanation, code fences, or quotes. Answer in English.{ANTI_ECHO}"
-            )
-        }
-        LocalGenerationProfile::Balanced => {
-            format!(
-                "You generate Git commit messages. Respond with a Conventional Commits subject line, then a blank line, then at most two short bullet points. Keep every line under 72 characters. No preamble, explanation, code fences, or quotes. Answer in English.{ANTI_ECHO}"
-            )
-        }
-        LocalGenerationProfile::Detailed => {
-            format!(
-                "You generate Git commit messages. Respond with a Conventional Commits subject line, then a blank line, then a concise body and up to four short bullet points grouped by affected area/file. Keep every line under 72 characters. No preamble, explanation, code fences, or quotes. Answer in English.{ANTI_ECHO}"
-            )
-        }
-    };
-
-    let mut user = String::new();
-    if let Some(n) = notes.filter(|n| !n.trim().is_empty()) {
-        user.push_str(&format!("Developer notes:\n{n}\n\n"));
-    }
-    user.push_str(&format!("Staged changes:\n{diff}"));
-    Ok((system, user))
-}
-
 pub(crate) fn build_messages(diff: &str, notes: Option<&str>) -> Result<(String, String), String> {
     if diff.trim().is_empty() {
         return Err("No staged changes available for a commit message.".to_string());
     }
 
-    // Rough token estimate — small models often have an 8-32k context window.
+    // Rough token estimate to keep requests within common context windows.
     const MAX_CHARS: usize = 24_000;
     let diff = truncate_at_char_boundary(diff, MAX_CHARS);
 

src-tauri/src/git.rs

@@ -578,6 +578,25 @@ fn repository_bundle_for_repo(
     // branches -> tags -> stashes -> commits -> files waterfall down to the
     // duration of its slowest member.
     let status = status_for_repo(repo)?;
+
+    // A freshly initialized or cloned empty repository has a symbolic HEAD,
+    // but it does not resolve to a commit yet (an "unborn" HEAD). Some Git
+    // commands and Git extensions treat that as a hard revision error. Keep
+    // the repository usable and still report any untracked working-tree files
+    // without starting commit-dependent workers.
+    if verify_commit(repo, "HEAD").is_err() {
+        let files = repository_files_with_status(repo, &status)?;
+        return Ok(RepositoryBundle {
+            status,
+            branches: Vec::new(),
+            tags: Vec::new(),
+            stashes: Vec::new(),
+            commits: Vec::new(),
+            files,
+            warning: None,
+        });
+    }
+
     let (branches, tags, stashes, commits, files) = thread::scope(|scope| -> Result<_, String> {
         let branches = scope.spawn(|| branches_for_repo(repo));
         let tags = scope.spawn(|| tags_for_repo(repo));
@@ -2117,28 +2136,6 @@ pub fn get_file_patch(path: String, file: String, staged: bool) -> Result<String
     Ok(String::from_utf8_lossy(&output).to_string())
 }
 
-#[tauri::command]
-pub fn commit_ai_local_models() -> Vec<commit_ai::LocalModelOption> {
-    commit_ai::LOCAL_MODELS.to_vec()
-}
-
-#[tauri::command]
-pub async fn commit_ai_status(
-    engine: tauri::State<'_, commit_ai::CommitAiEngine>,
-) -> Result<commit_ai::CommitAiStatus, String> {
-    Ok(engine.status().await)
-}
-
-/// Kicks off the (first-run-only) download and model load in the background and returns
-/// immediately; the frontend polls `commit_ai_status` to know when it's ready.
-#[tauri::command]
-pub fn commit_ai_load(model_id: String, engine: tauri::State<'_, commit_ai::CommitAiEngine>) {
-    let engine = engine.inner().clone();
-    tauri::async_runtime::spawn(async move {
-        engine.ensure_loaded(&model_id).await;
-    });
-}
-
 // The prompt is built from `git diff --cached` only, i.e. exactly the staged changes —
 // unstaged edits and untracked files never influence the generated message.
 fn staged_diff(repo: &Path) -> Result<String, String> {
@@ -2147,8 +2144,8 @@ fn staged_diff(repo: &Path) -> Result<String, String> {
     let name_status = run_git(repo, ["diff", "--cached", "--name-status", "-M"])?;
     let file_list = String::from_utf8_lossy(&name_status).trim().to_string();
 
-    // Generated lockfiles say nothing useful about intent but easily blow the small
+    // Generated lockfiles say little about intent and can easily dominate the context,
-    // context window of local models, so keep them out of the detailed diff.
+    // so keep them out of the detailed diff.
     let diff = run_git(
         repo,
         [
@@ -2178,50 +2175,6 @@ fn staged_diff(repo: &Path) -> Result<String, String> {
     Ok(format!("Staged files:\n{file_list}\n\n{diff}"))
 }
 
-fn staged_diff_local(
-    repo: &Path,
-    profile: commit_ai::LocalGenerationProfile,
-) -> Result<String, String> {
-    let name_status = run_git(repo, ["diff", "--cached", "--name-status", "-M"])?;
-    let file_list = String::from_utf8_lossy(&name_status).trim().to_string();
-
-    let stat = run_git(repo, ["diff", "--cached", "--stat", "--summary"])?;
-    let stat = String::from_utf8_lossy(&stat).trim().to_string();
-
-    let diff_args = vec![
-        "diff",
-        "--cached",
-        "--no-ext-diff",
-        "--no-textconv",
-        profile.diff_unified_context(),
-        "--",
-        ".",
-        ":(exclude)*package-lock.json",
-        ":(exclude)*pnpm-lock.yaml",
-        ":(exclude)*yarn.lock",
-        ":(exclude)*bun.lockb",
-        ":(exclude)*Cargo.lock",
-        ":(exclude)*composer.lock",
-        ":(exclude)*Gemfile.lock",
-        ":(exclude)*poetry.lock",
-        ":(exclude)*go.sum",
-    ];
-    let diff = run_git(repo, diff_args)?;
-    let diff = String::from_utf8_lossy(&diff).trim().to_string();
-
-    let mut sections = Vec::new();
-    if !file_list.is_empty() {
-        sections.push(format!("Staged files:\n{file_list}"));
-    }
-    if !stat.is_empty() {
-        sections.push(format!("Diff stat:\n{stat}"));
-    }
-    if !diff.is_empty() {
-        sections.push(format!("Detailed diff:\n{diff}"));
-    }
-    Ok(sections.join("\n\n"))
-}
-
 #[tauri::command]
 pub async fn commit_ai_generate(
     path: String,
@@ -2230,27 +2183,15 @@ pub async fn commit_ai_generate(
     model: Option<String>,
     api_key: Option<String>,
     base_url: Option<String>,
-    local_profile: Option<String>,
-    engine: tauri::State<'_, commit_ai::CommitAiEngine>,
 ) -> Result<String, String> {
     let repo = resolve_repo(&path)?;
-    let local_profile = commit_ai::LocalGenerationProfile::from_id(local_profile.as_deref());
+    let diff = staged_diff(&repo)?;
-    let diff = if provider == "local" {
-        staged_diff_local(&repo, local_profile)?
-    } else {
-        staged_diff(&repo)?
-    };
     let notes = notes.as_deref();
     let model = model.filter(|value| !value.trim().is_empty());
     let api_key = api_key.filter(|value| !value.trim().is_empty());
     let base_url = base_url.filter(|value| !value.trim().is_empty());
 
     match provider.as_str() {
-        "local" => {
-            engine
-                .generate_commit_message(&diff, notes, local_profile)
-                .await
-        }
         "openai" => {
             let api_key = api_key.ok_or_else(|| "OpenAI API key is missing.".to_string())?;
             let model = model.unwrap_or_else(|| "gpt-4o-mini".to_string());
@@ -2420,7 +2361,6 @@ pub async fn commit_ai_split(
             )
             .await?
         }
-        "local" => return Err("Commit splitting currently requires an API provider.".to_string()),
         other => return Err(format!("Unknown AI provider: {other}")),
     };
     parse_ai_commit_plan(&raw, &staged_files)
@@ -2522,7 +2462,6 @@ pub async fn commit_ai_review(
             let model = model.ok_or_else(|| "Model name is missing.".to_string())?;
             commit_ai::review_custom(&base_url, api_key.as_deref(), &model, &diff).await?
         }
-        "local" => return Err("Pre-commit review currently requires an API provider.".to_string()),
         other => return Err(format!("Unknown AI provider: {other}")),
     };
     parse_ai_review(&raw)
@@ -5481,19 +5420,26 @@ fn clone_repository_core(
     run_git_clone(remote_url.trim(), &target, username, password)?;
 
     let repo = resolve_repo(&target.to_string_lossy())?;
-    let lfs_warning = sync_git_lfs_objects_if_needed(
+    let lfs_warning = if verify_commit(&repo, "HEAD").is_ok() {
-        &repo,
+        sync_git_lfs_objects_if_needed(
-        Some("origin"),
+            &repo,
-        username,
+            Some("origin"),
-        password,
+            username,
-        true,
+            password,
-    )
+            true,
-    .err()
-    .map(|error| {
-        format!(
-            "Repository cloned, but Git LFS objects could not be downloaded automatically: {error}"
         )
-    });
+        .err()
+        .map(|error| {
+            format!(
+                "Repository cloned, but Git LFS objects could not be downloaded automatically: {error}"
+            )
+        })
+    } else {
+        // There cannot be LFS pointers to download before the first commit.
+        // In particular, avoid Git LFS implementations that try to resolve
+        // HEAD themselves and fail on an empty repository.
+        None
+    };
 
     let mut bundle = repository_bundle_for_repo(&repo, commit_limit)?;
     bundle.warning = lfs_warning;
@@ -7840,6 +7786,37 @@ mod tests {
         assert!(bundle.warning.is_none());
     }
 
+    #[test]
+    fn clone_repository_core_supports_empty_repository() {
+        let source = init_bare_temp_repo("empty_clone_source");
+        let parent = temp_dir("empty_clone_parent");
+
+        let bundle = clone_repository_core(
+            source.path.to_str().expect("source path should be UTF-8"),
+            parent.path.to_str().expect("parent path should be UTF-8"),
+            Some("local-copy"),
+            None,
+            None,
+            Some(100),
+        )
+        .expect("empty repository should clone");
+
+        let cloned_repo = parent.path.join("local-copy");
+        assert_eq!(
+            PathBuf::from(bundle.status.repo_path),
+            cloned_repo
+                .canonicalize()
+                .expect("clone path should resolve")
+        );
+        assert!(bundle.status.clean);
+        assert!(bundle.commits.is_empty());
+        assert!(bundle.branches.is_empty());
+        assert!(bundle.tags.is_empty());
+        assert!(bundle.stashes.is_empty());
+        assert!(bundle.files.is_empty());
+        assert!(bundle.warning.is_none());
+    }
+
     #[test]
     #[cfg_attr(
         windows,

src-tauri/src/main.rs

@@ -14,24 +14,23 @@ use git::{
     SearchCancellationState, add_remote, add_to_gitignore, add_worktree, amend_commit,
     apply_file_patch, cancel_code_search, cancel_file_history, checkout_branch, cherry_pick_abort,
     cherry_pick_commit, cherry_pick_continue, clone_repository, commit, commit_ai_generate,
-    commit_ai_load, commit_ai_local_models, commit_ai_review, commit_ai_split, commit_ai_status,
+    commit_ai_review, commit_ai_split, compare_commits, compare_file_to_head,
-    compare_commits, compare_file_to_head, compare_file_to_parent, create_branch, create_tag,
+    compare_file_to_parent, create_branch, create_tag, cred_delete, cred_load, cred_save,
-    cred_delete, cred_load, cred_save, delete_branch, delete_commit_note, delete_remote_branch,
+    delete_branch, delete_commit_note, delete_remote_branch, delete_remote_branches, delete_tag,
-    delete_remote_branches, delete_tag, diff_file_against_working_tree, fetch, fetch_commit_notes,
+    diff_file_against_working_tree, fetch, fetch_commit_notes, get_commit_note, get_file_blame,
-    get_commit_note, get_file_blame, get_file_patch, get_remote_url, get_status, git_lfs_install,
+    get_file_patch, get_remote_url, get_status, git_lfs_install, git_lfs_prune, git_lfs_pull,
-    git_lfs_prune, git_lfs_pull, git_lfs_status, git_lfs_track, git_lfs_untrack, init_repository,
+    git_lfs_status, git_lfs_track, git_lfs_untrack, init_repository, last_commit_message,
-    last_commit_message, list_branches, list_commits, list_file_history,
+    list_branches, list_commits, list_file_history, list_interactive_rebase_commits, list_reflog,
-    list_interactive_rebase_commits, list_reflog, list_remotes, list_repository_files,
+    list_remotes, list_repository_files, list_stashes, list_tags, list_worktrees, lock_worktree,
-    list_stashes, list_tags, list_worktrees, lock_worktree, merge_abort, merge_branch,
+    merge_abort, merge_branch, merge_continue, move_worktree, open_repo_in_explorer,
-    merge_continue, move_worktree, open_repo_in_explorer, open_repository, open_repository_bundle,
+    open_repository, open_repository_bundle, open_repository_file, prune_worktrees, pull, push,
-    open_repository_file, prune_worktrees, pull, push, push_commit_notes, push_tag, read_conflict,
+    push_commit_notes, push_tag, read_conflict, rebase_abort, rebase_branch, rebase_continue,
-    rebase_abort, rebase_branch, rebase_continue, remove_remote, remove_worktree, rename_branch,
+    remove_remote, remove_worktree, rename_branch, rename_remote_branch, repair_worktree,
-    rename_remote_branch, repair_worktree, resolve_conflict, resolve_conflict_side,
+    resolve_conflict, resolve_conflict_side, restore_file_from_commit, restore_files,
-    restore_file_from_commit, restore_files, restore_reflog_entry, restore_to_commit,
+    restore_reflog_entry, restore_to_commit, revert_commit, run_sequence_editor_if_requested,
-    revert_commit, run_sequence_editor_if_requested, search_code_introductions,
+    search_code_introductions, set_branch_upstream, set_commit_note, stage_files,
-    set_branch_upstream, set_commit_note, stage_files, start_interactive_rebase, stash_apply,
+    start_interactive_rebase, stash_apply, stash_drop, stash_pop, stash_push, undo_last_commit,
-    stash_drop, stash_pop, stash_push, undo_last_commit, unlock_worktree, unstage_files,
+    unlock_worktree, unstage_files, untrack_paths, update_remote,
-    untrack_paths, update_remote,
 };
 use integrations::list_integration_repositories;
 use std::path::{Path, PathBuf};
@@ -317,7 +316,6 @@ async fn main() {
         .manage(StartupRepository(Mutex::new(startup_repository)))
         .manage(StartupClone(Mutex::new(startup_clone)))
         .manage(SearchCancellationState::default())
-        .manage(commit_ai::CommitAiEngine::new())
         .plugin(tauri_plugin_dialog::init());
 
     // Linux installs are expected to come from the system package manager (see the
@@ -388,9 +386,6 @@ async fn main() {
             amend_commit,
             undo_last_commit,
             last_commit_message,
-            commit_ai_status,
-            commit_ai_load,
-            commit_ai_local_models,
             commit_ai_generate,
             commit_ai_review,
             commit_ai_split,

src/App.svelte

@@ -49,9 +49,6 @@
     commitAiGenerate,
     commitAiReview,
     commitAiSplit,
-    commitAiLoad,
-    commitAiLocalModels,
-    commitAiStatus,
     compareCommits,
     cancelCodeSearch,
     cancelFileHistory,
@@ -151,7 +148,6 @@
     AppLanguage,
     AppTheme,
     AnalyticsSettings,
-    CommitAiPhase,
     CustomThemeColors,
     ConflictFile,
     DetectedExternalTool,
@@ -179,7 +175,6 @@
     GitStatus,
     GitTag,
     GitWorktree,
-    LocalModelOption,
     PatchApplyAction,
     PreparedResolution,
     RebaseCommit,
@@ -360,8 +355,6 @@
   let commitMessage = "";
   let amendMode = false;
   let preAmendDraftMessage = "";
-  let lastLocalAiGeneratedMessage = "";
-  let commitAiPhase: CommitAiPhase = "idle";
   let commitAiGenerating = false;
   let commitAiReviewing = false;
   let commitAiSplitting = false;
@@ -369,7 +362,6 @@
   let aiCommitSplitOpen = false;
   let aiReviewResult: AiReviewResult | null = null;
   let aiReviewOpen = false;
-  let commitAiPollTimer: ReturnType<typeof setInterval> | undefined;
   let aiSettings: AiSettings = defaultAiSettings();
   let aiSettingsOpen = false;
   let appSettingsOpen = false;
@@ -387,7 +379,6 @@
   let detectedExternalTools: DetectedExternalTool[] = [];
   let externalToolsDetectionPending = true;
   let externalToolsDetectionUnavailable = false;
-  let localModelOptions: LocalModelOption[] = [];
   let errorMessage = "";
   let operation = "";
   let compareFrom = "";
@@ -628,7 +619,6 @@
     if (autoRefreshTimer) { clearInterval(autoRefreshTimer); autoRefreshTimer = undefined; }
     if (backgroundRepoStatusTimer) { clearInterval(backgroundRepoStatusTimer); backgroundRepoStatusTimer = undefined; }
     if (backgroundFetchTimer) { clearInterval(backgroundFetchTimer); backgroundFetchTimer = undefined; }
-    if (commitAiPollTimer) { clearInterval(commitAiPollTimer); commitAiPollTimer = undefined; }
   }
 
   function wait(ms: number): Promise<void> {
@@ -681,7 +671,7 @@
     loadRepoLists();
 
     void checkForUpdates();
-    void initCommitAi();
+    aiSettings = loadAiSettings();
 
     try {
       await waitForStartupPaint();
@@ -1029,48 +1019,10 @@
 
   // ── Commit AI ──────────────────────────────────────────────────────────────
 
-  function stopCommitAiPolling() {
-    if (commitAiPollTimer) { clearInterval(commitAiPollTimer); commitAiPollTimer = undefined; }
-  }
-
-  async function pollCommitAiStatus() {
-    try {
-      const result = await commitAiStatus();
-      commitAiPhase = result.phase;
-    } catch { /* ignore transient errors */ }
-    if (commitAiPhase === "ready" || commitAiPhase === "error") stopCommitAiPolling();
-  }
-
-  function startCommitAiPolling() {
-    // Only the local model has a download/load phase worth polling — cloud providers are
-    // plain API calls with nothing to wait for.
-    stopCommitAiPolling();
-    if (aiSettings.provider !== "local") return;
-    void pollCommitAiStatus();
-    commitAiPollTimer = setInterval(() => { void pollCommitAiStatus(); }, 2000);
-  }
-
-  async function initCommitAi() {
-    aiSettings = loadAiSettings();
-    try {
-      localModelOptions = await commitAiLocalModels();
-    } catch { /* AI features stay disabled if this fails; not fatal to the app */ }
-    if (aiSettings.provider === "local") {
-      try { await commitAiLoad(aiSettings.localModelId); } catch { /* surfaced via status polling */ }
-    }
-    startCommitAiPolling();
-  }
-
   function saveAiSettings(next: AiSettings) {
-    const modelChanged = next.provider === "local" && next.localModelId !== aiSettings.localModelId;
     aiSettings = next;
     persistAiSettings(next);
     aiSettingsOpen = false;
-    if (next.provider === "local" && (modelChanged || commitAiPhase === "idle")) {
-      commitAiPhase = "idle";
-      void commitAiLoad(next.localModelId);
-    }
-    startCommitAiPolling();
   }
 
   function defaultAnalyticsSettings(): AnalyticsSettings {
@@ -1324,27 +1276,15 @@
 
   function updateCommitMessage(message: string) {
     commitMessage = message;
-    if (message !== lastLocalAiGeneratedMessage) {
-      lastLocalAiGeneratedMessage = "";
-    }
   }
 
   async function generateCommitMessageWithAi() {
     if (!activeRepoPath || commitAiGenerating) return;
-    if (aiSettings.provider === "local" && commitAiPhase !== "ready") return;
     commitAiGenerating = true;
     errorMessage = "";
     try {
       const notes = commitMessage.trim() || undefined;
-      if (aiSettings.provider === "local") {
+      if (aiSettings.provider === "openai") {
-        const localNotes = notes && notes !== lastLocalAiGeneratedMessage ? notes : undefined;
-        commitMessage = await commitAiGenerate(activeRepoPath, {
-          provider: "local",
-          notes: localNotes,
-          localProfile: aiSettings.localProfile,
-        });
-        lastLocalAiGeneratedMessage = commitMessage;
-      } else if (aiSettings.provider === "openai") {
         const cred = await credLoad("ai:openai");
         commitMessage = await commitAiGenerate(activeRepoPath, {
           provider: "openai",
@@ -1352,7 +1292,6 @@
           model: aiSettings.openaiModel,
           apiKey: cred?.password,
         });
-        lastLocalAiGeneratedMessage = "";
       } else if (aiSettings.provider === "anthropic") {
         const cred = await credLoad("ai:anthropic");
         commitMessage = await commitAiGenerate(activeRepoPath, {
@@ -1361,7 +1300,6 @@
           model: aiSettings.anthropicModel,
           apiKey: cred?.password,
         });
-        lastLocalAiGeneratedMessage = "";
       } else {
         const cred = await credLoad("ai:custom");
         commitMessage = await commitAiGenerate(activeRepoPath, {
@@ -1371,7 +1309,6 @@
           baseUrl: aiSettings.customBaseUrl,
           apiKey: cred?.password,
         });
-        lastLocalAiGeneratedMessage = "";
       }
     } catch (error) {
       errorMessage = errorToMessage(error);
@@ -1382,10 +1319,6 @@
 
   async function reviewStagedWithAi() {
     if (!activeRepoPath || commitAiReviewing || stagedCount === 0) return;
-    if (aiSettings.provider === "local") {
-      errorMessage = "Pre-commit review currently requires OpenAI, Anthropic, or a custom endpoint.";
-      return;
-    }
     commitAiReviewing = true;
     errorMessage = "";
     try {
@@ -1427,10 +1360,6 @@
 
   async function splitStagedWithAi() {
     if (!activeRepoPath || commitAiSplitting || stagedCount < 2) return;
-    if (aiSettings.provider === "local") {
-      errorMessage = "Commit splitting currently requires OpenAI, Anthropic, or a custom endpoint.";
-      return;
-    }
     commitAiSplitting = true;
     errorMessage = "";
     try {
@@ -1735,8 +1664,6 @@
   function defaultAiSettings(): AiSettings {
     return {
       provider: "openai",
-      localModelId: "qwen2.5-0.5b",
-      localProfile: "fast",
       openaiModel: "gpt-4o-mini",
       anthropicModel: "claude-3-5-haiku-latest",
       customBaseUrl: "",
@@ -1748,11 +1675,17 @@
     try {
       const stored = JSON.parse(localStorage.getItem(AI_SETTINGS_KEY) ?? "null") as unknown;
       if (stored && typeof stored === "object") {
-        const merged = { ...defaultAiSettings(), ...(stored as Partial<AiSettings>) };
+        const candidate = stored as Partial<AiSettings> & { provider?: unknown };
-        // Local AI is still in development and disabled in the settings UI — migrate any
+        const provider = candidate.provider === "anthropic" || candidate.provider === "custom"
-        // previously saved selection away from it so nobody gets stuck on a dead option.
+          ? candidate.provider
-        if (merged.provider === "local") merged.provider = "openai";
+          : "openai";
-        return merged;
+        return {
+          provider,
+          openaiModel: typeof candidate.openaiModel === "string" ? candidate.openaiModel : "gpt-4o-mini",
+          anthropicModel: typeof candidate.anthropicModel === "string" ? candidate.anthropicModel : "claude-3-5-haiku-latest",
+          customBaseUrl: typeof candidate.customBaseUrl === "string" ? candidate.customBaseUrl : "",
+          customModel: typeof candidate.customModel === "string" ? candidate.customModel : "",
+        };
       }
     } catch {
       // Fall through to defaults below.
@@ -4385,7 +4318,6 @@
         commitMessage = "";
         amendMode = false;
         preAmendDraftMessage = "";
-        lastLocalAiGeneratedMessage = "";
         await refreshRepositoryViews(activeRepoPath);
         trackEvent("commit_created", { amend: 1 });
       });
@@ -4396,7 +4328,6 @@
     await runOperation("Committing", async () => {
       applyStatus(await commit(activeRepoPath, message));
       commitMessage = "";
-      lastLocalAiGeneratedMessage = "";
       await refreshRepositoryViews(activeRepoPath);
       trackEvent("commit_created", { amend: 0, staged_files: trackedStagedCount });
     });
@@ -5617,8 +5548,6 @@
             {isBusy}
             {operation}
             {stagedCount}
-            commitAiProvider={aiSettings.provider}
-            {commitAiPhase}
             {commitAiGenerating}
             {commitAiReviewing}
             {commitAiSplitting}
@@ -5983,7 +5912,6 @@
   {#await import("./lib/components/AiSettingsDialog.svelte") then module}
     <module.default
       settings={aiSettings}
-      localModels={localModelOptions}
       onSave={saveAiSettings}
       onClose={() => { aiSettingsOpen = false; }}
     />

src/app.css

@@ -4160,44 +4160,6 @@
     color: #f5f7ff;
     background: linear-gradient(180deg, rgba(100,108,255,0.22), rgba(65,209,255,0.1));
   }
-  .ai-provider-option-local {
-    flex-wrap: wrap;
-    row-gap: 2px;
-  }
-  .ai-provider-badge {
-    flex-basis: 100%;
-    text-align: center;
-    font-size: 9.5px;
-    font-weight: 700;
-    letter-spacing: 0.04em;
-    text-transform: uppercase;
-    color: var(--color-ink-faint);
-  }
-  .ai-local-profile-options {
-    display: grid;
-    grid-template-columns: repeat(3, minmax(0, 1fr));
-    gap: 8px;
-  }
-  .ai-local-profile-option {
-    min-width: 0;
-    min-height: 34px;
-    padding: 0 8px;
-    border-color: var(--color-border-subtle);
-    background: rgba(255,255,255,0.03);
-    color: var(--color-ink-dim);
-    font-size: 12px;
-    font-weight: 700;
-  }
-  .ai-local-profile-option:hover:not(:disabled) {
-    border-color: var(--color-border);
-    color: var(--color-ink);
-    background: var(--color-surface-hover);
-  }
-  .ai-local-profile-option.active {
-    border-color: rgba(65,209,255,0.48);
-    color: #f5f7ff;
-    background: linear-gradient(180deg, rgba(65,209,255,0.16), rgba(100,108,255,0.12));
-  }
   .new-branch-form {
     display: flex;
     flex-direction: column;

src/lib/components/AiReviewDialog.svelte

@@ -15,8 +15,7 @@
   function providerLabel(value: CommitAiProvider): string {
     if (value === "openai") return "OpenAI";
     if (value === "anthropic") return "Anthropic";
-    if (value === "custom") return "Custom endpoint";
+    return "Custom endpoint";
-    return "Local AI";
   }
 
   function locationLabel(finding: AiReviewFinding): string {

src/lib/components/AiSettingsDialog.svelte

@@ -1,20 +1,18 @@
 <script lang="ts">
   import { onDestroy, onMount } from "svelte";
-  import { AlertCircle, Bot, Check, Cpu, Eye, EyeOff, Gauge, Globe, Key, LoaderCircle, Sparkles, X, Zap } from "@lucide/svelte";
+  import { Bot, Check, Eye, EyeOff, Globe, Key, LoaderCircle, X } from "@lucide/svelte";
   import { credDelete, credLoad, credSave } from "../git";
-  import type { AiSettings, CommitAiLocalProfile, CommitAiProvider, LocalModelOption } from "../types";
+  import type { AiSettings, CommitAiProvider } from "../types";
-  import SelectMenu from "./SelectMenu.svelte";
 
   interface Props {
     settings: AiSettings;
-    localModels: LocalModelOption[];
     onSave: (settings: AiSettings) => void;
     onClose: () => void;
   }
 
-  let { settings, localModels = [], onSave, onClose }: Props = $props();
+  let { settings, onSave, onClose }: Props = $props();
 
-  type CloudProvider = Exclude<CommitAiProvider, "local">;
+  type CloudProvider = CommitAiProvider;
 
   const CRED_KEYS: Record<CloudProvider, string> = {
     openai: "ai:openai",
@@ -22,9 +20,7 @@
     custom: "ai:custom",
   };
 
-  let provider = $state<CommitAiProvider>("local");
+  let provider = $state<CommitAiProvider>("openai");
-  let localModelId = $state("");
-  let localProfile = $state<CommitAiLocalProfile>("fast");
   let openaiModel = $state("");
   let anthropicModel = $state("");
   let customBaseUrl = $state("");
@@ -41,8 +37,6 @@
 
   $effect(() => {
     provider = settings.provider;
-    localModelId = settings.localModelId;
-    localProfile = settings.localProfile ?? "fast";
     openaiModel = settings.openaiModel;
     anthropicModel = settings.anthropicModel;
     customBaseUrl = settings.customBaseUrl;
@@ -103,8 +97,6 @@
       ]);
       onSave({
         provider,
-        localModelId,
-        localProfile,
         openaiModel: openaiModel.trim() || "gpt-4o-mini",
         anthropicModel: anthropicModel.trim() || "claude-3-5-haiku-latest",
         customBaseUrl: customBaseUrl.trim(),
@@ -117,26 +109,6 @@
     }
   }
 
-  function formatSize(mb: number): string {
-    return mb >= 1000 ? `${(mb / 1000).toFixed(1)} GB` : `${mb} MB`;
-  }
-
-  function recommendedModelForProfile(profile: CommitAiLocalProfile): string {
-    if (profile === "balanced") return "qwen2.5-1.5b";
-    if (profile === "detailed") return "qwen2.5-3b";
-    return "qwen2.5-0.5b";
-  }
-
-  function selectLocalProfile(profile: CommitAiLocalProfile) {
-    const previousRecommended = recommendedModelForProfile(localProfile);
-    localProfile = profile;
-    const nextRecommended = recommendedModelForProfile(profile);
-    if (!localModelId || localModelId === previousRecommended) {
-      localModelId = nextRecommended;
-    }
-  }
-
-  let selectedLocalModel = $derived(localModels.find((option) => option.id === localModelId));
 </script>
 
 <div
@@ -156,17 +128,6 @@
 
     <form class="ai-settings-form" onsubmit={(e) => { e.preventDefault(); void handleSave(); }}>
       <div class="ai-provider-options" role="radiogroup" aria-label="AI provider">
-        <button
-          type="button"
-          class="ai-provider-option ai-provider-option-local"
-          class:active={provider === "local"}
-          disabled
-          title="Local AI is still in development and not yet available"
-        >
-          <Cpu size={16} aria-hidden="true" />
-          Local AI
-          <span class="ai-provider-badge">In development</span>
-        </button>
         <button type="button" class="ai-provider-option" class:active={provider === "openai"} onclick={() => { provider = "openai"; }}>
           <Bot size={16} aria-hidden="true" />
           OpenAI
@@ -181,38 +142,7 @@
         </button>
       </div>
 
-      {#if provider === "local"}
+      {#if provider === "openai"}
-        <div class="cred-field">
-          <span class="cred-field-label">Local speed</span>
-          <div class="ai-local-profile-options" role="radiogroup" aria-label="Local AI speed">
-            <button type="button" class="ai-local-profile-option" class:active={localProfile === "fast"} onclick={() => selectLocalProfile("fast")}>
-              <Zap size={15} aria-hidden="true" />
-              Fast
-            </button>
-            <button type="button" class="ai-local-profile-option" class:active={localProfile === "balanced"} onclick={() => selectLocalProfile("balanced")}>
-              <Gauge size={15} aria-hidden="true" />
-              Balanced
-            </button>
-            <button type="button" class="ai-local-profile-option" class:active={localProfile === "detailed"} onclick={() => selectLocalProfile("detailed")}>
-              <Sparkles size={15} aria-hidden="true" />
-              Detailed
-            </button>
-          </div>
-        </div>
-        <label class="cred-field">
-          <span class="cred-field-label">Model</span>
-          <SelectMenu value={localModelId} options={localModels.map((option) => ({ value: option.id, label: `${option.label} - ${formatSize(option.approx_size_mb)}` }))} onChange={(value) => { localModelId = value; }} />
-        </label>
-        <div class="cred-token-hint">
-          <AlertCircle size={13} aria-hidden="true" />
-          <span>
-            Switching downloads the model{selectedLocalModel ? ` (${formatSize(selectedLocalModel.approx_size_mb)})` : ""}
-            in the background — depending on your internet connection this can take several minutes.
-            After that it stays cached locally and loads instantly on the next start.
-            The speed setting only changes Local AI; API providers keep their existing prompt.
-          </span>
-        </div>
-      {:else if provider === "openai"}
         <label class="cred-field">
           <span class="cred-field-label">Model</span>
           <input type="text" bind:value={openaiModel} placeholder="gpt-4o-mini" autocomplete="off" spellcheck="false" />

src/lib/components/CommitPanel.svelte

@@ -1,6 +1,5 @@
 <script lang="ts">
   import { Check, GitCommitHorizontal, LoaderCircle, RotateCcw, Settings, ShieldCheck, Sparkles } from "@lucide/svelte";
-  import type { CommitAiPhase, CommitAiProvider } from "../types";
 
   interface Props {
     commitMessage: string;
@@ -10,8 +9,6 @@
     isBusy: boolean;
     operation: string;
     stagedCount: number;
-    commitAiProvider: CommitAiProvider;
-    commitAiPhase: CommitAiPhase;
     commitAiGenerating: boolean;
     commitAiReviewing: boolean;
     commitAiSplitting: boolean;
@@ -35,8 +32,6 @@
     isBusy = false,
     operation = "",
     stagedCount = 0,
-    commitAiProvider = "local",
-    commitAiPhase = "idle",
     commitAiGenerating = false,
     commitAiReviewing = false,
     commitAiSplitting = false,
@@ -57,23 +52,19 @@
     onCommit();
   }
 
-  function aiButtonTitle(provider: CommitAiProvider, phase: CommitAiPhase, staged: number): string {
+  function aiButtonTitle(staged: number): string {
     if (staged === 0) return "Stage changes first";
-    if (provider === "local" && phase === "loading") return "AI model is downloading/loading — this happens once";
-    if (provider === "local" && phase === "error") return "AI model failed to load — check AI settings";
     return "Generate commit message with AI from the staged diff";
   }
 
-  let localModelLoading = $derived(commitAiProvider === "local" && commitAiPhase === "loading");
   let canGenerate = $derived(
     hasRepository &&
     !isBusy &&
     !commitAiGenerating &&
     !commitAiReviewing &&
-    stagedCount > 0 &&
+    stagedCount > 0,
-    (commitAiProvider !== "local" || commitAiPhase === "ready"),
   );
-  let canReview = $derived(canGenerate && commitAiProvider !== "local");
+  let canReview = $derived(canGenerate);
   let canSplit = $derived(canReview && stagedCount > 1 && !commitAiSplitting);
 </script>
 
@@ -90,7 +81,7 @@
         type="button"
         onclick={onSplitStaged}
         disabled={!canSplit}
-        title={commitAiProvider === "local" ? "Commit splitting currently requires an API provider" : "Suggest logical commits for the staged files"}
+        title="Suggest logical commits for the staged files"
       >
         {#if commitAiSplitting}<LoaderCircle class="spin" size={14} aria-hidden="true" />{:else}<GitCommitHorizontal size={14} aria-hidden="true" />{/if}
         Split
@@ -100,7 +91,7 @@
         type="button"
         onclick={onReviewStaged}
         disabled={!canReview}
-        title={commitAiProvider === "local" ? "Pre-commit review currently requires an API provider" : stagedCount === 0 ? "Stage changes first" : "Review staged changes for bugs and risks"}
+        title={stagedCount === 0 ? "Stage changes first" : "Review staged changes for bugs and risks"}
       >
         {#if commitAiReviewing}<LoaderCircle class="spin" size={14} aria-hidden="true" />{:else}<ShieldCheck size={14} aria-hidden="true" />{/if}
         Review
@@ -110,9 +101,9 @@
         type="button"
         onclick={onGenerateCommitMessage}
         disabled={!canGenerate}
-        title={aiButtonTitle(commitAiProvider, commitAiPhase, stagedCount)}
+        title={aiButtonTitle(stagedCount)}
       >
-        {#if commitAiGenerating || localModelLoading}<LoaderCircle class="spin" size={14} aria-hidden="true" />{:else}<Sparkles size={14} aria-hidden="true" />{/if}
+        {#if commitAiGenerating}<LoaderCircle class="spin" size={14} aria-hidden="true" />{:else}<Sparkles size={14} aria-hidden="true" />{/if}
         Generate
       </button>
       <button class="commit-settings-button" type="button" onclick={onOpenAiSettings} disabled={isBusy} title="AI settings" aria-label="AI settings">

src/lib/git.ts

@@ -3,9 +3,7 @@ import { tracedInvoke as invoke } from "./telemetry";
 import type {
   AiReviewResult,
   AiCommitPlan,
-  CommitAiLocalProfile,
   CommitAiProvider,
-  CommitAiStatus,
   ConflictFile,
   DetectedExternalTool,
   ExternalToolCommand,
@@ -30,7 +28,6 @@ import type {
   GitStatus,
   GitTag,
   GitWorktree,
-  LocalModelOption,
   PatchApplyAction,
   RepositoryBundle,
   StoredCredential,
@@ -369,22 +366,9 @@ export function stashDrop(path: string, selector: string): Promise<GitStatus> {
   return invoke<GitStatus>("stash_drop", { path, selector });
 }
 
-export function commitAiStatus(): Promise<CommitAiStatus> {
-  return invoke<CommitAiStatus>("commit_ai_status");
-}
-
-export function commitAiLoad(modelId: string): Promise<void> {
-  return invoke<void>("commit_ai_load", { modelId });
-}
-
-export function commitAiLocalModels(): Promise<LocalModelOption[]> {
-  return invoke<LocalModelOption[]>("commit_ai_local_models");
-}
-
 export interface CommitAiGenerateOptions {
   notes?: string;
   provider: CommitAiProvider;
-  localProfile?: CommitAiLocalProfile;
   model?: string;
   apiKey?: string;
   baseUrl?: string;
@@ -395,7 +379,6 @@ export function commitAiGenerate(path: string, options: CommitAiGenerateOptions)
     path,
     notes: options.notes,
     provider: options.provider,
-    localProfile: options.localProfile,
     model: options.model,
     apiKey: options.apiKey,
     baseUrl: options.baseUrl,

src/lib/types.ts

@@ -9,9 +9,7 @@ export type FileStatusKind =
 
 export type GitIgnoreKind = "file" | "extension" | "folder";
 
-export type CommitAiPhase = "idle" | "loading" | "ready" | "error";
+export type CommitAiProvider = "openai" | "anthropic" | "custom";
-export type CommitAiProvider = "local" | "openai" | "anthropic" | "custom";
-export type CommitAiLocalProfile = "fast" | "balanced" | "detailed";
 export type AppTheme = "system" | "light" | "dark";
 export type AppAppearance = "modern" | "classic" | "custom";
 export type AppLanguage = "en" | "de";
@@ -74,12 +72,6 @@ export interface CustomThemeColors {
   text: string;
 }
 
-export interface CommitAiStatus {
-  phase: CommitAiPhase;
-  model_id: string | null;
-  error: string | null;
-}
-
 export type AiReviewRisk = "low" | "medium" | "high";
 export type AiReviewSeverity = "critical" | "warning" | "info";
 
@@ -109,16 +101,8 @@ export interface AiCommitPlan {
   groups: AiCommitGroup[];
 }
 
-export interface LocalModelOption {
-  id: string;
-  label: string;
-  approx_size_mb: number;
-}
-
 export interface AiSettings {
   provider: CommitAiProvider;
-  localModelId: string;
-  localProfile: CommitAiLocalProfile;
   openaiModel: string;
   anthropicModel: string;
   customBaseUrl: string;