CI: disclose truncated repository file trees (block/buzz#4499)

desktop/playwright.config.ts

@@ -105,6 +105,7 @@ export default defineConfig({
         "**/inbox-edit.spec.ts",
         "**/send-channel-binding.spec.ts",
         "**/project-commit-detail.spec.ts",
+        "**/project-file-tree.spec.ts",
         "**/project-inbox.spec.ts",
         "**/project-issue-comments.spec.ts",
         "**/project-pr-review.spec.ts",

desktop/src-tauri/src/commands/mod.rs

@@ -47,6 +47,7 @@ mod project_git;
 mod project_git_branches;
 mod project_git_diff;
 mod project_git_exec;
+mod project_git_files;
 mod project_git_merge_error;
 mod project_git_push;
 mod project_git_workflow;

desktop/src-tauri/src/commands/project_git.rs

@@ -2,11 +2,11 @@ use super::project_git_exec::{
     build_git_auth_config, clean_branch, clean_target_ref, run_git, validate_workspace_clone_url,
     GitAuthConfig,
 };
+use super::project_git_files::{parse_ls_tree, parse_worktree_files, ParsedProjectRepoFiles};
 use super::project_git_push::push_project_local_repository_blocking;
 use super::project_repo_paths::{canonical_repos_roots, find_local_repo_dir};
 use crate::app_state::AppState;
 use serde::Serialize;
-use std::time::UNIX_EPOCH;
 use tauri::State;
 #[derive(Clone, Serialize)]
 pub struct ProjectRepoCommitInfo {
@@ -38,6 +38,8 @@ pub struct ProjectRepoSnapshotInfo {
     pub latest_commit: Option<ProjectRepoCommitInfo>,
     pub commits: Vec<ProjectRepoCommitInfo>,
     pub files: Vec<ProjectRepoFileInfo>,
+    /// Complete file count before the snapshot payload is capped.
+    pub total_file_count: usize,
     pub contributors: Vec<ProjectRepoContributorInfo>,
 }
 #[derive(Serialize)]
@@ -134,30 +136,6 @@ fn has_untracked_files(output: &str) -> bool {
     output.lines().any(|line| line.starts_with("??"))
 }
 
-fn read_preview_content(
-    repo_dir: &std::path::Path,
-    path: &str,
-    size: Option<u64>,
-) -> Option<String> {
-    const MAX_PREVIEW_BYTES: u64 = 64 * 1024;
-    if size.is_some_and(|value| value > MAX_PREVIEW_BYTES) {
-        return None;
-    }
-
-    let full_path = repo_dir.join(path);
-    let normalized = full_path.canonicalize().ok()?;
-    let repo_root = repo_dir.canonicalize().ok()?;
-    if !normalized.starts_with(repo_root) {
-        return None;
-    }
-
-    let bytes = std::fs::read(normalized).ok()?;
-    if bytes.contains(&0) {
-        return None;
-    }
-    String::from_utf8(bytes).ok()
-}
-
 fn parse_commits(output: &str) -> Vec<ProjectRepoCommitInfo> {
     output
         .lines()
@@ -235,46 +213,6 @@ fn parse_latest_commit_by_path(
     result
 }
 
-fn path_modified_at(path: &std::path::Path) -> Option<i64> {
-    let modified = std::fs::metadata(path).ok()?.modified().ok()?;
-    modified
-        .duration_since(UNIX_EPOCH)
-        .ok()
-        .and_then(|duration| i64::try_from(duration.as_secs()).ok())
-}
-
-fn parse_worktree_files(
-    repo_dir: &std::path::Path,
-    output: &str,
-    latest_commit_by_path: &std::collections::HashMap<String, ProjectRepoCommitInfo>,
-) -> Vec<ProjectRepoFileInfo> {
-    output
-        .split('\0')
-        .filter(|path| !path.trim().is_empty())
-        .filter_map(|path| {
-            let full_path = repo_dir.join(path);
-            let metadata = std::fs::metadata(&full_path).ok()?;
-            if !metadata.is_file() {
-                return None;
-            }
-            let size = Some(metadata.len());
-            let latest_commit = latest_commit_by_path.get(path).cloned();
-            Some(ProjectRepoFileInfo {
-                path: path.to_string(),
-                kind: "blob".to_string(),
-                size,
-                preview_content: read_preview_content(repo_dir, path, size),
-                last_changed_at: latest_commit
-                    .as_ref()
-                    .map(|commit| commit.timestamp)
-                    .or_else(|| path_modified_at(&full_path)),
-                latest_commit,
-            })
-        })
-        .take(250)
-        .collect()
-}
-
 fn normalize_branch_name(branch: &str) -> &str {
     branch
         .trim()
@@ -309,40 +247,6 @@ fn branch_activity_range(
     Some(format!("origin/{base_branch}..HEAD"))
 }
 
-fn parse_ls_tree(
-    repo_dir: &std::path::Path,
-    output: &str,
-    latest_commit_by_path: &std::collections::HashMap<String, ProjectRepoCommitInfo>,
-) -> Vec<ProjectRepoFileInfo> {
-    output
-        .lines()
-        .filter_map(|line| {
-            let (meta, path) = line.split_once('\t')?;
-            let mut parts = meta.split_whitespace();
-            let _mode = parts.next()?;
-            let kind = parts.next()?.to_string();
-            let _object = parts.next()?;
-            let size = parts.next().and_then(|value| value.parse::<u64>().ok());
-            let preview_content = if kind == "blob" {
-                read_preview_content(repo_dir, path, size)
-            } else {
-                None
-            };
-            Some(ProjectRepoFileInfo {
-                path: path.to_string(),
-                kind,
-                size,
-                preview_content,
-                last_changed_at: latest_commit_by_path
-                    .get(path)
-                    .map(|commit| commit.timestamp),
-                latest_commit: latest_commit_by_path.get(path).cloned(),
-            })
-        })
-        .take(250)
-        .collect()
-}
-
 fn snapshot_from_repo(
     repo_dir: &std::path::Path,
     auth: &GitAuthConfig,
@@ -383,7 +287,7 @@ fn snapshot_from_repo(
         (Vec::new(), Vec::new())
     };
 
-    let files = if latest_commit.is_some() {
+    let parsed_files = if latest_commit.is_some() {
         let latest_commit_by_path = run_git(
             &[
                 "log",
@@ -402,13 +306,14 @@ fn snapshot_from_repo(
             .map(|output| parse_ls_tree(repo_dir, &output, &latest_commit_by_path))
             .unwrap_or_default()
     } else {
-        Vec::new()
+        ParsedProjectRepoFiles::default()
     };
 
     ProjectRepoSnapshotInfo {
         latest_commit,
         commits,
-        files,
+        files: parsed_files.files,
+        total_file_count: parsed_files.total_file_count,
         contributors,
     }
 }
@@ -466,7 +371,7 @@ fn snapshot_from_worktree(
         (Vec::new(), Vec::new(), std::collections::HashMap::new())
     };
 
-    let files = run_git(
+    let parsed_files = run_git(
         &[
             "ls-files",
             "--cached",
@@ -483,7 +388,8 @@ fn snapshot_from_worktree(
     ProjectRepoSnapshotInfo {
         latest_commit,
         commits,
-        files,
+        files: parsed_files.files,
+        total_file_count: parsed_files.total_file_count,
         contributors,
     }
 }

desktop/src-tauri/src/commands/project_git_files.rs

@@ -0,0 +1,152 @@
+use super::project_git::{ProjectRepoCommitInfo, ProjectRepoFileInfo};
+use std::{collections::HashMap, path::Path, time::UNIX_EPOCH};
+
+const MAX_REPOSITORY_FILE_PAYLOAD: usize = 250;
+
+#[derive(Default)]
+pub(super) struct ParsedProjectRepoFiles {
+    pub(super) files: Vec<ProjectRepoFileInfo>,
+    pub(super) total_file_count: usize,
+}
+
+fn read_preview_content(repo_dir: &Path, path: &str, size: Option<u64>) -> Option<String> {
+    const MAX_PREVIEW_BYTES: u64 = 64 * 1024;
+    if size.is_some_and(|value| value > MAX_PREVIEW_BYTES) {
+        return None;
+    }
+
+    let normalized = repo_dir.join(path).canonicalize().ok()?;
+    let repo_root = repo_dir.canonicalize().ok()?;
+    if !normalized.starts_with(repo_root) {
+        return None;
+    }
+
+    let bytes = std::fs::read(normalized).ok()?;
+    if bytes.contains(&0) {
+        return None;
+    }
+    String::from_utf8(bytes).ok()
+}
+
+fn path_modified_at(path: &Path) -> Option<i64> {
+    let modified = std::fs::metadata(path).ok()?.modified().ok()?;
+    modified
+        .duration_since(UNIX_EPOCH)
+        .ok()
+        .and_then(|duration| i64::try_from(duration.as_secs()).ok())
+}
+
+pub(super) fn parse_worktree_files(
+    repo_dir: &Path,
+    output: &str,
+    latest_commit_by_path: &HashMap<String, ProjectRepoCommitInfo>,
+) -> ParsedProjectRepoFiles {
+    let mut parsed = ParsedProjectRepoFiles::default();
+
+    for path in output.split('\0').filter(|path| !path.trim().is_empty()) {
+        let full_path = repo_dir.join(path);
+        let Ok(metadata) = std::fs::metadata(&full_path) else {
+            continue;
+        };
+        if !metadata.is_file() {
+            continue;
+        }
+
+        parsed.total_file_count += 1;
+        if parsed.files.len() >= MAX_REPOSITORY_FILE_PAYLOAD {
+            continue;
+        }
+
+        let size = Some(metadata.len());
+        let latest_commit = latest_commit_by_path.get(path).cloned();
+        parsed.files.push(ProjectRepoFileInfo {
+            path: path.to_string(),
+            kind: "blob".to_string(),
+            size,
+            preview_content: read_preview_content(repo_dir, path, size),
+            last_changed_at: latest_commit
+                .as_ref()
+                .map(|commit| commit.timestamp)
+                .or_else(|| path_modified_at(&full_path)),
+            latest_commit,
+        });
+    }
+
+    parsed
+}
+
+pub(super) fn parse_ls_tree(
+    repo_dir: &Path,
+    output: &str,
+    latest_commit_by_path: &HashMap<String, ProjectRepoCommitInfo>,
+) -> ParsedProjectRepoFiles {
+    let mut parsed = ParsedProjectRepoFiles::default();
+
+    for line in output.lines() {
+        let Some((meta, path)) = line.split_once('\t') else {
+            continue;
+        };
+        let mut parts = meta.split_whitespace();
+        let (Some(_mode), Some(kind), Some(_object)) = (parts.next(), parts.next(), parts.next())
+        else {
+            continue;
+        };
+        let size = parts.next().and_then(|value| value.parse::<u64>().ok());
+
+        parsed.total_file_count += 1;
+        if parsed.files.len() >= MAX_REPOSITORY_FILE_PAYLOAD {
+            continue;
+        }
+
+        let preview_content = (kind == "blob")
+            .then(|| read_preview_content(repo_dir, path, size))
+            .flatten();
+        parsed.files.push(ProjectRepoFileInfo {
+            path: path.to_string(),
+            kind: kind.to_string(),
+            size,
+            preview_content,
+            last_changed_at: latest_commit_by_path
+                .get(path)
+                .map(|commit| commit.timestamp),
+            latest_commit: latest_commit_by_path.get(path).cloned(),
+        });
+    }
+
+    parsed
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn ls_tree_reports_the_full_count_when_the_payload_is_capped() {
+        let output = (0..251)
+            .map(|index| format!("100644 blob {} 1\tfile-{index:03}.txt", "0".repeat(40)))
+            .collect::<Vec<_>>()
+            .join("\n");
+
+        let parsed = parse_ls_tree(Path::new("/missing-repository"), &output, &HashMap::new());
+
+        assert_eq!(parsed.files.len(), 250);
+        assert_eq!(parsed.total_file_count, 251);
+    }
+
+    #[test]
+    fn worktree_reports_the_full_count_when_the_payload_is_capped() {
+        let repo = tempfile::tempdir().expect("create temporary repository");
+        let mut paths = Vec::new();
+        for index in 0..251 {
+            let path = format!("file-{index:03}.txt");
+            std::fs::write(repo.path().join(&path), "x").expect("write fixture file");
+            paths.push(path);
+        }
+        let output = format!("{}\0", paths.join("\0"));
+
+        let parsed = parse_worktree_files(repo.path(), &output, &HashMap::new());
+
+        assert_eq!(parsed.files.len(), 250);
+        assert_eq!(parsed.total_file_count, 251);
+    }
+}

desktop/src/features/agents/AGENTS.md

@@ -113,13 +113,16 @@ with a TypeScript lookup table or an id comparison in a component.
    Once the Advanced toggle is visible, its expanded state is exclusively
    user-controlled: provider, harness, and required-env changes must never
    open it automatically in defaults, create, or edit flows. In Create mode,
-   the defaults summary follows preferred-harness changes saved while the
-   dialog is open, and its configured state includes required credentials as
-   well as provider/model values. If no available harness can resolve, Create
-   starts in Customize and lets unavailable catalog entries be selected only
-   to expose their setup guidance; submission remains blocked.
-   Advanced-only required credentials mark the collapsed Advanced toggle
-   without opening it in Global Defaults and Edit, and block incomplete saves.
+   `Run on` belongs in Advanced directly after **Who can send instructions**;
+   keep it out of the basic create fields. The defaults summary follows
+   preferred-harness changes saved while the dialog is open, and its configured
+   state includes required credentials as well as provider/model values. If no
+   available harness can resolve, Create starts in Customize and lets unavailable
+   catalog entries be selected only to expose their setup guidance; submission
+   remains blocked.
+   Advanced-only required credentials and incomplete remote **Run on** setup
+   mark the collapsed Advanced toggle without opening it, and block incomplete
+   saves.
    Runtime-file credentials satisfy Global Defaults just as they do Create and
    Edit. In Edit,
    selecting Custom command keeps its required command field beside the harness