batch3 T4: file_ops + file_index services; UI endpoints; tools refactor

- services/file_ops.ts: shared listDir/viewFile/grep/findFiles core
- services/file_index.ts: per-project flat file list cached on mtime of
  project root + .git/HEAD + .git/index (rg --files honors .gitignore)
- services/tools.ts: tools delegate to file_ops, output format unchanged
- routes/projects.ts: GET /list_dir, /view_file, /files endpoints
- web client: api.projects.listDir/viewFile/files + mirrored types

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

apps/server/src/services/tools.ts

@@ -3,6 +3,7 @@ import { resolve, basename, relative } from 'node:path';
 import { spawn } from 'node:child_process';
 import { z } from 'zod';
 import { pathGuard, PathScopeError } from './path_guard.js';
+import { grep as fileOpsGrep } from './file_ops.js';
 
 const MAX_FILE_BYTES = 5 * 1024 * 1024;
 const DEFAULT_VIEW_LINES = 200;
@@ -168,15 +169,6 @@ const GrepInput = z.object({
 });
 type GrepInputT = z.infer<typeof GrepInput>;
 
-interface RipgrepMatch {
-  type: string;
-  data?: {
-    path?: { text?: string };
-    line_number?: number;
-    lines?: { text?: string };
-  };
-}
-
 export const grep: ToolDef<GrepInputT> = {
   name: 'grep',
   description:
@@ -203,73 +195,27 @@ export const grep: ToolDef<GrepInputT> = {
     },
   },
   async execute(input, projectRoot) {
-    const target = await pathGuard(projectRoot, input.path ?? projectRoot);
     const limit = Math.min(
       Math.max(input.max_results ?? DEFAULT_GREP_RESULTS, 1),
       MAX_GREP_RESULTS
     );
-    const args = [
-      '--json',
-      '--max-count',
-      String(limit),
-      '--max-columns',
-      '300',
-    ];
-    if (!input.case_sensitive) args.push('--ignore-case');
-    if (input.hidden) args.push('--hidden');
-    args.push('--', input.pattern, target);
-
-    return await new Promise((resolveP, rejectP) => {
-      const child = spawn('rg', args, { cwd: projectRoot });
-      const matches: Array<{ path: string; line: number; content: string }> = [];
-      let buf = '';
-      let stderr = '';
-      child.stdout.setEncoding('utf8');
-      child.stderr.setEncoding('utf8');
-      child.stdout.on('data', (chunk: string) => {
-        buf += chunk;
-        let idx;
-        while ((idx = buf.indexOf('\n')) >= 0) {
-          const line = buf.slice(0, idx);
-          buf = buf.slice(idx + 1);
-          if (!line) continue;
-          if (matches.length >= limit) continue;
-          try {
-            const parsed = JSON.parse(line) as RipgrepMatch;
-            if (parsed.type !== 'match' || !parsed.data) continue;
-            const path = parsed.data.path?.text ?? '';
-            const lineNumber = parsed.data.line_number ?? 0;
-            const content = parsed.data.lines?.text ?? '';
-            matches.push({
-              path: relative(projectRoot, path) || path,
-              line: lineNumber,
-              content: content.replace(/\n$/, ''),
-            });
-          } catch {
-            /* ignore non-json */
-          }
-        }
-        if (matches.length >= limit) {
-          child.kill();
-        }
-      });
-      child.stderr.on('data', (chunk: string) => {
-        stderr += chunk;
-      });
-      child.on('error', (err) => rejectP(err));
-      child.on('close', (code) => {
-        // rg exits 1 when no matches, 2 on real error
-        if (code === 2 && matches.length === 0) {
-          rejectP(new Error(`ripgrep failed: ${stderr.slice(0, 300)}`));
-          return;
-        }
-        resolveP({
-          matches,
-          total: matches.length,
-          truncated: matches.length >= limit,
-        });
-      });
+    // Delegate to file_ops.grep; reshape match objects to preserve LLM output format
+    // (file_ops uses {path, line, text}; tool output uses {path, line, content})
+    const result = await fileOpsGrep(projectRoot, input.pattern, {
+      path: input.path,
+      max_matches: limit,
+      case_sensitive: input.case_sensitive,
+      hidden: input.hidden,
     });
+    return {
+      matches: result.matches.map((m) => ({
+        path: m.path,
+        line: m.line,
+        content: m.text,
+      })),
+      total: result.matches.length,
+      truncated: result.truncated,
+    };
   },
 };