beast-agent 1.7.0 → 1.8.0

package/src/agent/tools.js

@@ -1,2273 +1,2340 @@
-'use strict';
-
-const fs = require('fs');
-const os = require('os');
-const path = require('path');
-const https = require('https');
-const { execFile } = require('child_process');
-const { spawn } = require('child_process');
-const research = require('./research');
-const obscura = require('./obscura');
-
-const MAX_CMD_OUTPUT = 16000;
-const MAX_FILE_CHARS = 200000;
-
-function truncateMiddle(s, cap) {
-  if (s.length <= cap) return s;
-  const half = Math.floor((cap - 32) / 2);
-  return (
-    s.slice(0, half) +
-    `\n... [${s.length - cap} chars truncated] ...\n` +
-    s.slice(s.length - half)
-  );
-}
-
-/* ---------- opencode port: truncate.ts ----------
-   MAX_LINES 2000 / MAX_BYTES 50KB aşarsa TAM çıktı geçici dosyaya yazılır;
-   modele sınırlı önizleme + dosya yolu döner (Model Tool Output bounding).
-   Böylece hiçbir araç çıktısı KAYBOLMAZ — ajan read_file ile kalanını okur. */
-const TRUNC_MAX_LINES = 2000;
-const TRUNC_MAX_BYTES = 50 * 1024;
-const TRUNC_PREVIEW_CHARS = 6000; /* engine'in 7200'lik dilimi içinde ipucu kalsın */
-
-function toolOutputDir() {
-  const d = path.join(os.tmpdir(), 'beast-tool-output');
-  try { fs.mkdirSync(d, { recursive: true }); } catch {}
-  return d;
-}
-
-function boundToolOutput(text) {
-  const s = String(text || '');
-  const bytes = Buffer.byteLength(s, 'utf8');
-  const lines = s.split('\n').length;
-  if (bytes <= TRUNC_MAX_BYTES && lines <= TRUNC_MAX_LINES) return { text: s };
-  const file = path.join(
-    toolOutputDir(),
-    `tool_${Date.now().toString(36)}_${Math.random().toString(36).slice(2, 7)}.txt`
-  );
-  try {
-    fs.writeFileSync(file, s);
-  } catch {
-    return { text: s.slice(0, TRUNC_PREVIEW_CHARS) + '\n…[kırpıldı]' };
-  }
-  return {
-    text:
-      s.slice(0, TRUNC_PREVIEW_CHARS) +
-      `\n\n[çıktı kırpıldı: ${lines} satır / ${bytes} byte — TAM ÇIKTI: ${file}\nread_file'ı offset/limit ile kullanarak kalan bölümleri oku]`,
-    outputFile: file,
-  };
-}
-
-/* ---------- opencode port: shell.ts + shell/prompt.ts ----------
-   Workspace başına KALICI PowerShell oturumu: cd/env çağrılar arasında
-   korunur, her çağrıda process spawn maliyeti yoktur. Komut bitişi benzersiz
-   işaretle algılanır (prompt senkronizasyonu). Paralel çağrılar oturum
-   başına sıraya girer (opencode ile aynı: oturum başına sıralı yürütme). */
-const SHELL_QUEUE_CAP = 6; /* en fazla bu kadar ayrı workspace oturumu */
-const SHELL_IDLE_MS = 10 * 60 * 1000; /* 10 dk boşta kalan oturum kapatılır */
-const _shSessions = new Map(); // cwd → session
-
-/* boşta reaper: unref'li zamanlayıcı — event loop'u tek başına tutmaz */
-const _shellReaper = setInterval(() => {
-  const now = Date.now();
-  for (const [k, sess] of _shSessions) {
-    if (!sess.busy && now - (sess.lastUsed || 0) > SHELL_IDLE_MS) {
-      _shellDispose(sess);
-      _shSessions.delete(k);
-    }
-  }
-}, 60000);
-if (_shellReaper.unref) _shellReaper.unref();
-
-function _shellDispose(sess) {
-  try {
-    if (sess.proc && sess.proc.pid) {
-      spawn('taskkill', ['/pid', String(sess.proc.pid), '/T', '/F'], { windowsHide: true });
-    }
-  } catch {}
-  try { sess.proc && sess.proc.kill(); } catch {}
-  sess.dead = true;
-}
-
-function disposeShellSessions() {
-  for (const sess of _shSessions.values()) _shellDispose(sess);
-  _shSessions.clear();
-}
-
-function _spawnShell(cwd) {
-  const proc = spawn(
-    'powershell.exe',
-    ['-NoProfile', '-NoExit', '-Command', '-'],
-    { cwd: cwd || process.cwd(), windowsHide: true, env: envWithPathPrefix() }
-  );
-  const sess = {
-    proc,
-    cwd,
-    buf: '',
-    err: '',
-    seq: 0,
-    busy: false,
-    queue: [], // {command, resolve, timer, signal, onAbort}
-    dead: false,
-  };
-  /* Node 22: child + stdio stream'leri event loop'a bağlanmaz — test/CLI'da
-     bekleyen oturum sürecin çıkmasını ENGELLEMEZ; uygulama kapanışında
-     disposeShellSessions() yine de temiz kapatır */
-  try {
-    proc.unref && proc.unref();
-    proc.stdin.unref && proc.stdin.unref();
-    proc.stdout.unref && proc.stdout.unref();
-    proc.stderr.unref && proc.stderr.unref();
-  } catch {}
-  proc.stdin.on && proc.stdin.on('error', () => {}); // EPIPE yut — oturum ölünce yazma patlamasın
-  proc.stdout.setEncoding('utf8');
-  proc.stderr.setEncoding('utf8');
-  proc.stdout.on('data', (d) => {
-    sess.buf += d;
-    _shellPump(sess);
-  });
-  proc.stderr.on('data', (d) => {
-    sess.err += d;
-    if (sess.err.length > MAX_CMD_OUTPUT * 4) sess.err = sess.err.slice(-MAX_CMD_OUTPUT * 2);
-  });
-  proc.on('exit', () => {
-    sess.dead = true;
-    /* bekleyen komutları ölü oturumda düşür — yeni çağrı taze oturum açar */
-    while (sess.queue.length) {
-      const w = sess.queue.shift();
-      clearTimeout(w.timer);
-      w.resolve({ ok: false, code: null, output: sess.buf.slice(-2000) + '\n[beast] shell oturumu kapandı' });
-    }
-    sess.busy = false;
-    if (_shSessions.get(cwd) === sess) _shSessions.delete(cwd);
-  });
-  proc.on('error', () => {
-    sess.dead = true;
-  });
-  return sess;
-}
-
-function _shellPump(sess) {
-  const w = sess.queue[0];
-  if (!w || !w.dispatched) return;
-  const marker = `${SHELL_MARKER_PREFIX}${w.n}_`;
-  const mi = sess.buf.indexOf(marker);
-  if (mi < 0) return;
-  const lineEnd = sess.buf.indexOf('\n', mi);
-  if (lineEnd < 0) return; /* işaret satırı tam gelmedi — daha fazla veri bekle */
-  const markerLine = sess.buf.slice(mi, lineEnd).trim();
-  let out = sess.buf.slice(0, mi);
-  sess.buf = sess.buf.slice(lineEnd + 1);
-  sess.queue.shift();
-  sess.busy = sess.queue.length > 0;
-  clearTimeout(w.timer);
-  const codeM = /_(\-?\d+)\s*$/.exec(markerLine);
-  const exitCode = codeM ? Number(codeM[1]) : null;
-  if (out.length > MAX_CMD_OUTPUT * 4) out = out.slice(-MAX_CMD_OUTPUT * 2);
-  const errPart = sess.err.trim();
-  sess.err = '';
-  w.resolve({
-    ok: exitCode === 0,
-    code: exitCode,
-    output:
-      (out.trim() || '') +
-      (errPart ? (out.trim() ? '\n[stderr] ' : '') + errPart : ''),
-  });
-  sess.lastUsed = Date.now();
-  if (sess.queue.length) _shellDispatch(sess);
-}
-
-const SHELL_MARKER_PREFIX = '__BEAST_DONE_';
-
-function _shellDispatch(sess) {
-  const w = sess.queue[0];
-  if (!w || w.dispatched) return;
-  w.dispatched = true;
-  w.n = ++sess.seq;
-  const cmd = w.command;
-  const marker = `${SHELL_MARKER_PREFIX}${w.n}_$LASTEXITCODE`;
-  /* $LASTEXITCODE sıfırlanır: yalnız cmdlet koşan komutlar da ok:true dönsün;
-     native exe hata verirse gerçek kod işaret satırına yazılır */
-  sess.proc.stdin.write(`$global:LASTEXITCODE = 0\n${cmd}${cmd.endsWith('\n') ? '' : '\n'}Write-Output "${marker}"\n`);
-  const finishTimeout = () => {
-    _shellDispose(sess);
-    if (_shSessions.get(sess.cwd) === sess) _shSessions.delete(sess.cwd);
-    sess.busy = false;
-    while (sess.queue.length) {
-      const x = sess.queue.shift();
-      clearTimeout(x.timer);
-      x.resolve({ ok: false, code: null, output: '[beast] shell komutu zaman aşımı — oturum tazelendi' });
-    }
-  };
-  w.timer = setTimeout(finishTimeout, w.timeoutMs);
-  if (w.signal) {
-    if (w.signal.aborted) return finishTimeout();
-    w.signal.addEventListener('abort', finishTimeout, { once: true });
-  }
-}
-
-function runShellCommand(command, cwd, timeoutMs = 90000, signal) {
-  return new Promise((resolve) => {
-    let sess = _shSessions.get(cwd);
-    if (!sess || sess.dead) {
-      try {
-        sess = _spawnShell(cwd);
-        _shSessions.set(cwd, sess);
-        while (_shSessions.size > SHELL_QUEUE_CAP) {
-          const [k, old] = _shSessions.entries().next().value;
-          if (k === cwd) break;
-          _shellDispose(old);
-          _shSessions.delete(k);
-        }
-      } catch {
-        /* kalıcı oturum açılamadı → tek seferlik klasik yol */
-        return runCommand(command, cwd, timeoutMs, signal).then(resolve);
-      }
-    }
-    sess.queue.push({ command, resolve, timeoutMs, signal, dispatched: false });
-    sess.lastUsed = Date.now();
-    if (!sess.busy) {
-      sess.busy = true;
-      _shellDispatch(sess);
-    }
-  });
-}
-
-/* --- core/shell.ts gitbash() port: Windows'ta bash çözümleme sırası --- */
-function gitbash() {
-  const cands = [];
-  if (process.env.ProgramFiles) cands.push(path.join(process.env.ProgramFiles, 'Git', 'bin', 'bash.exe'));
-  if (process.env['ProgramFiles(x86)']) cands.push(path.join(process.env['ProgramFiles(x86)'], 'Git', 'bin', 'bash.exe'));
-  if (process.env.LOCALAPPDATA) cands.push(path.join(process.env.LOCALAPPDATA, 'Programs', 'Git', 'bin', 'bash.exe'));
-  for (const c of cands) {
-    try { if (fs.existsSync(c)) return c; } catch {}
-  }
-  return 'bash'; /* PATH'te aranır; yoksa spawn ENOENT → net hata mesajı */
-}
-
-function runBashCommand(command, cwd, timeoutMs = 90000, signal) {
-  return new Promise((resolve) => {
-    let settled = false;
-    let out = '';
-    let err = '';
-    const finish = (code, killed) => {
-      if (settled) return;
-      settled = true;
-      clearTimeout(timer);
-      const text = (out ? out.trim() : '') + (err ? (out ? '\n[stderr] ' : '') + err.trim() : '');
-      resolve({ ok: !killed && code === 0, code, output: truncateMiddle(text || '(no output)', MAX_CMD_OUTPUT) });
-    };
-    let child;
-    try {
-      child = spawn(gitbash(), ['-lc', command], {
-        cwd: cwd || process.cwd(),
-        windowsHide: true,
-        env: envWithPathPrefix(),
-      });
-    } catch {
-      resolve({ ok: false, code: null, output: 'bash bulunamadı — Git for Windows kur ya da PowerShell kullan' });
-      return;
-    }
-    const timer = setTimeout(() => {
-      try { spawn('taskkill', ['/pid', String(child.pid), '/T', '/F'], { windowsHide: true }); } catch {}
-      err += '\n[beast] command timed out';
-      finish(null, true);
-    }, timeoutMs);
-    if (signal) {
-      if (signal.aborted) return finish(null, true);
-      signal.addEventListener('abort', () => {
-        try { spawn('taskkill', ['/pid', String(child.pid), '/T', '/F'], { windowsHide: true }); } catch {}
-        finish(null, true);
-      }, { once: true });
-    }
-    child.stdout.setEncoding('utf8');
-    child.stderr.setEncoding('utf8');
-    child.stdout.on('data', (d) => { out += d; });
-    child.stderr.on('data', (d) => { err += d; });
-    child.on('error', (e) => {
-      clearTimeout(timer);
-      settled = true;
-      const msg = String((e && e.message) || '');
-      resolve({
-        ok: false,
-        code: null,
-        output: /ENOENT/i.test(msg)
-          ? 'bash bulunamadı — Git for Windows kur ya da PowerShell kullan'
-          : msg,
-      });
-    });
-    child.on('close', (code) => finish(code, false));
-  });
-}
-
-function runCommand(command, cwd, timeoutMs = 90000, signal) {
-  return new Promise((resolve) => {
-    let settled = false;
-    let out = '';
-    let err = '';
-    const finish = (code, killed) => {
-      if (settled) return;
-      settled = true;
-      clearTimeout(timer);
-      const text =
-        (out ? out.trim() : '') +
-        (err ? (out ? '\n[stderr] ' : '') + err.trim() : '');
-      resolve({
-        ok: !killed && code === 0,
-        output: truncateMiddle(text || '(no output)', MAX_CMD_OUTPUT),
-        code,
-      });
-    };
-
-    const child = spawn(
-      'powershell.exe',
-      ['-NoProfile', '-NonInteractive', '-ExecutionPolicy', 'Bypass', '-Command', command],
-      { cwd: cwd || process.cwd(), windowsHide: true, env: envWithPathPrefix() }
-    );
-
-    const timer = setTimeout(() => {
-      try {
-        spawn('taskkill', ['/pid', String(child.pid), '/T', '/F'], { windowsHide: true });
-      } catch {}
-      finish(null, true);
-      // append notice
-      err += '\n[beast] command timed out';
-    }, timeoutMs);
-
-    if (signal) {
-      if (signal.aborted) return finish(null, true);
-      signal.addEventListener('abort', () => {
-        try {
-          spawn('taskkill', ['/pid', String(child.pid), '/T', '/F'], { windowsHide: true });
-        } catch {}
-        finish(null, true);
-      }, { once: true });
-    }
-
-    child.stdout.setEncoding('utf8');
-    child.stderr.setEncoding('utf8');
-    child.stdout.on('data', (d) => {
-      out += d;
-      if (out.length > MAX_CMD_OUTPUT * 4) out = out.slice(-MAX_CMD_OUTPUT * 2);
-    });
-    child.stderr.on('data', (d) => {
-      err += d;
-      if (err.length > MAX_CMD_OUTPUT * 4) err = err.slice(-MAX_CMD_OUTPUT * 2);
-    });
-    child.on('error', (e) => finish(1, false) || void (err += String(e.message)));
-    child.on('close', (code) => finish(code, false));
-  });
-}
-
-function safeResolve(p, cwd) {
-  const abs = path.isAbsolute(p) ? p : path.join(cwd || process.cwd(), p);
-  return path.normalize(abs);
-}
-
-/* ---------- opencode read.ts port: binary tespit + fuzzy öneri ----------
-   opencode read.ts kuralı: uzantı kara listesi YA DA ilk 4KB'ta NUL byte YA DA
-   %30+'ı yazdırılamaz karakter → dosya binary sayılır, okunmaz. */
-const BINARY_EXT_RE =
-  /\.(zip|tar|gz|tgz|bz2|xz|7z|rar|exe|dll|so|dylib|bin|iso|img|msi|apk|jar|class|pyc|pyo|o|obj|a|lib|woff2?|ttf|otf|eot|mp3|mp4|avi|mkv|mov|flac|ogg|wav|webm|psd|ai|sketch|db|sqlite3?|pdb|docx?|xlsx?|pptx?|odt|ods|odp|pgp|gpg|keystore|jks|p12|traineddata|idx)$/i;
-
-function looksBinary(buf) {
-  const sample = buf.length > 4096 ? buf.subarray(0, 4096) : buf;
-  if (sample.includes(0)) return true;
-  let nonPrintable = 0;
-  for (const b of sample) {
-    if (b === 9 || b === 10 || b === 13) continue; // \t \n \r
-    if (b < 32 || b === 127 || b >= 0x80) nonPrintable++; // 0x80+ UTF-8 devam byte'ı olabilir ama kaba tarama yeterli
-  }
-  return sample.length > 0 && nonPrintable / sample.length > 0.3;
-}
-
-/* opencode read.ts: dosya yoksa aynı klasörde isim ön-ek benzerliği olan 3 kardeş öner */
-function fuzzySiblings(abs) {
-  try {
-    const dir = path.dirname(abs);
-    const base = path.basename(abs).toLowerCase();
-    const prefix = base.slice(0, 4);
-    const score = (name) => {
-      const n = name.toLowerCase();
-      if (n === base) return -1;
-      let s = 0;
-      if (n.startsWith(prefix)) s += 2;
-      for (let i = 0; i < Math.min(base.length, n.length); i++) if (base[i] === n[i]) s += 0.1;
-      return s;
-    };
-    return fs
-      .readdirSync(dir)
-      .map((name) => ({ name, s: score(name) }))
-      .filter((x) => x.s > 0)
-      .sort((a, b) => b.s - a.s)
-      .slice(0, 3)
-      .map((x) => x.name);
-  } catch {
-    return [];
-  }
-}
-
-/* ---------- opencode port: grep/glob altyapısı ----------
-   ripgrep gitignore'u izler; Beast'te bağımlılık olmasın diye ağır klasörler
-   statik atlanır (node_modules, .git, derleme çıktıları…). */
-const SKIP_DIRS = new Set([
-  'node_modules', '.git', '__pycache__', '.venv', 'venv', 'env',
-  'dist', 'build', 'out', 'coverage', '.next', '.turbo', '.cache',
-]);
-
-/* mini glob → regex: ** → her şey, * → / içermeyen her şey, ? → tek karakter */
-function globToRegExp(glob) {
-  const g = String(glob || '').trim();
-  if (!g) return null;
-  let re = '';
-  for (let i = 0; i < g.length; i++) {
-    const c = g[i];
-    if (c === '*') {
-      if (g[i + 1] === '*') {
-        re += '.*';
-        i++;
-        if (g[i + 1] === '/') i++; /* yıldız-yıldız-slash: üst klasörler opsiyonel */
-      } else re += '[^/\\\\]*';
-    } else if (c === '?') re += '[^/\\\\]';
-    else re += c.replace(/[.+^${}()|[\]\\]/g, '\\$&');
-  }
-  try {
-    return new RegExp('^' + re + '$', 'i');
-  } catch {
-    return null;
-  }
-}
-
-/* Sıralı dosya yürüyüşü; cb true dönerse erken durur */
-function walkFiles(dir, includeRe, cb, depth = 0) {
-  if (depth > 24) return false;
-  let entries = [];
-  try {
-    entries = fs.readdirSync(dir, { withFileTypes: true });
-  } catch {
-    return false;
-  }
-  for (const e of entries) {
-    const full = path.join(dir, e.name);
-    if (e.isDirectory()) {
-      if (SKIP_DIRS.has(e.name) || e.name.startsWith('.') && e.name !== '.opencode') continue;
-      if (walkFiles(full, includeRe, cb, depth + 1)) return true;
-    } else if (e.isFile()) {
-      if (includeRe && !includeRe.test(e.name)) continue;
-      if (cb(full)) return true;
-    }
-  }
-  return false;
-}
-
-/* ---------- Python altyapısı (#18) ----------
-   Sıra: BEAST_PYTHON env → sistem python/python3/py -3 → gömülü runtime
-   (%APPDATA%\beast\py — yoksa TEK SEFERLİK kendisi indirip açar).
-   Böylece ajanlar makinede Python kurulmasa bile python_run kullanabilir. */
-
-const PYTHON_EMBED_URL = 'https://www.python.org/ftp/python/3.12.8/python-3.12.8-embed-amd64.zip';
-const PY_PROBE_TIMEOUT = 6000;
-
-function beastAppDir() {
-  const base = process.env.APPDATA || path.join(os.homedir(), 'AppData', 'Roaming');
-  return path.join(base, 'beast');
-}
-
-function pythonScriptsDir() {
-  const d = path.join(beastAppDir(), 'scripts');
-  fs.mkdirSync(d, { recursive: true });
-  return d;
-}
-
-function embeddedPythonExe() {
-  return path.join(beastAppDir(), 'py', 'python.exe');
-}
-
-/* Gömülü runtime kuruluysa klasörünü PATH önüne koy: ajanın run_command'ında
-   `python` / `python.exe` da çalışsın (makinede Python kurulmuş olmasa bile). */
-function pythonPathPrefix() {
-  try {
-    const d = path.dirname(embeddedPythonExe());
-    if (fs.existsSync(path.join(d, 'python.exe'))) return d + path.delimiter;
-  } catch {}
-  return '';
-}
-
-function envWithPathPrefix() {
-  const base = { ...process.env };
-  base.PATH = pythonPathPrefix() + (process.env.PATH || process.env.Path || '');
-  return base;
-}
-
-/* paketlenmiş python scriptleri (src/agent/scripts) → kullanıcı scripts klasörüne tohumla */
-function bundledScriptPath(name) {
-  return path.join(__dirname, 'scripts', name);
-}
-
-function seedScript(name, force = false) {
-  const src = bundledScriptPath(name);
-  try {
-    const dest = path.join(pythonScriptsDir(), name);
-    /* force: paket içindeki güncel sürüm her zaman kazanır (sistem scripti) */
-    if (fs.existsSync(src) && (force || !fs.existsSync(dest))) {
-      fs.copyFileSync(src, dest);
-    }
-    return dest;
-  } catch {
-    return path.join(pythonScriptsDir(), name);
-  }
-}
-
-/* #20 hızlı web arama: ddgs kütüphanesi varsa onu kullan; yoksa python çoklu-motor
-   paralel; o da olmazsa JS fallback. Sonuç boş gelirse (CAPTCHA/engel) ddgs'yi
-   TEK SEFERLİK arka planda kurup tekrar dener. */
-function pyExec(exe, args, timeoutMs) {
-  return new Promise((resolve) => {
-    try {
-      execFile(exe, args, { timeout: timeoutMs, windowsHide: true }, (err, stdout, stderr) =>
-        resolve({ err, out: String(stdout || '') + String(stderr || '') })
-      );
-    } catch (e) {
-      resolve({ err: e, out: '' });
-    }
-  });
-}
-
-let _ddgsTried = null; // Promise<boolean> — oturum başına tek kurulum denemesi
-
-function ddgsImportOk(probe) {
-  /* hızlı yerel kontrol: kütüphane import edilebiliyor mu (ağ gerektirmez) */
-  return pyExec(probe.exe, ['-c', 'import ddgs'], 15000).then((r) => !r.err);
-}
-
-/* pip ile TEK SEFERLİK kurulum dener; uzun sürebilir — await ETME, arka planda bırak */
-function ensureDdgs(probe) {
-  if (_ddgsTried) return _ddgsTried;
-  _ddgsTried = (async () => {
-    try {
-      if (await ddgsImportOk(probe)) return true;
-      const pip = await pyExec(probe.exe, ['-m', 'pip', '--version'], 15000);
-      if (pip.err) return false; // gömülü python'da pip yok — sessizce vazgeç
-      const inst = await pyExec(
-        probe.exe,
-        ['-m', 'pip', 'install', '--quiet', '--disable-pip-version-check', 'ddgs'],
-        120000
-      );
-      if (inst.err) return false;
-      return await ddgsImportOk(probe);
-    } catch {
-      return false;
-    }
-  })();
-  return _ddgsTried;
-}
-
-function parseWsOutput(output) {
-  return String(output || '')
-    .split('\n')
-    .map((l) => l.trim())
-    .filter((l) => l.startsWith('{'))
-    .map((l) => {
-      try { return JSON.parse(l); } catch { return null; }
-    })
-    .filter(Boolean)
-    .map((x) => ({ title: String(x.title || ''), url: String(x.url || ''), snippet: String(x.snippet || ''), engine: String(x.engine || '') }));
-}
-
-async function webSearchFast(query, { maxResults = 8, signal } = {}) {
-  try {
-    const probe = await ensurePython({ allowDownload: true, signal });
-    const script = seedScript('websearch.py', true);
-    let r = await runPy(probe.exe, script, ['--json', '--limit', String(maxResults), query], process.cwd(), 20000, signal);
-    let results = parseWsOutput(r.output);
-    if (!results.length) {
-      /* muhtemel CAPTCHA/engel — ddgs kuruluysa hemen onunla tekrar dene;
-         kurulu değilse kurulumu arka plana bırak (bu çağrıyı bekletmez) */
-      if (await ddgsImportOk(probe)) {
-        r = await runPy(probe.exe, script, ['--json', '--limit', String(maxResults), query], process.cwd(), 20000, signal);
-        results = parseWsOutput(r.output);
-      } else {
-        ensureDdgs(probe).catch(() => {});
-      }
-    } else if (!results.some((x) => x.engine === 'ddgs')) {
-      /* arama çalışıyor ama ddgs henüz yok — arka planda sessiz kur, sonraki aramalar hızlanır */
-      ensureDdgs(probe).catch(() => {});
-    }
-    if (results.length) {
-      return { ok: true, engine: results.some((x) => x.engine === 'ddgs') ? 'ddgs' : 'python-multi', query, results };
-    }
-  } catch {}
-  /* JS fallback (tek motor DDG) */
-  return webSearch(query, { maxResults, signal });
-}
-
-let _pyProbe = null;
-
-function probeInterpreter(exe, prefix) {
-  return new Promise((resolve) => {
-    try {
-      execFile(exe, [...(prefix || []), '--version'], { timeout: PY_PROBE_TIMEOUT, windowsHide: true }, (err, stdout, stderr) => {
-        if (err) return resolve(null);
-        const v = String(stdout || stderr || '').trim();
-        resolve(/Python\s*3/i.test(v) ? { exe, prefix: prefix || [], source: 'system', version: v } : null);
-      });
-    } catch {
-      resolve(null);
-    }
-  });
-}
-
-async function findSystemPython() {
-  if (process.env.BEAST_PYTHON && fs.existsSync(process.env.BEAST_PYTHON)) {
-    return { exe: process.env.BEAST_PYTHON, prefix: [], source: 'env' };
-  }
-  for (const cand of [
-    ['python.exe', []],
-    ['python3.exe', []],
-    ['py.exe', ['-3']],
-  ]) {
-    const r = await probeInterpreter(cand[0], cand[1]);
-    if (r) return r;
-  }
-  return null;
-}
-
-function httpsGetBuffer(url, redirectsLeft = 4, signal) {
-  return new Promise((resolve, reject) => {
-    const req = https.get(url, { headers: { 'User-Agent': 'BeastAgent/1.0 (+python bootstrap)' } }, (res) => {
-      if ([301, 302, 303, 307, 308].includes(res.statusCode) && res.headers.location && redirectsLeft > 0) {
-        res.resume();
-        return resolve(httpsGetBuffer(new URL(res.headers.location, url).toString(), redirectsLeft - 1, signal));
-      }
-      if (res.statusCode !== 200) {
-        res.resume();
-        return reject(new Error(`indirme başarısız: HTTP ${res.statusCode}`));
-      }
-      const chunks = [];
-      let size = 0;
-      res.on('data', (c) => {
-        size += c.length;
-        if (size > 80 * 1024 * 1024) {
-          req.destroy(new Error('dosya çok büyük'));
-          return;
-        }
-        chunks.push(c);
-      });
-      res.on('end', () => resolve(Buffer.concat(chunks)));
-      res.on('error', reject);
-    });
-    req.on('error', reject);
-    if (signal) {
-      if (signal.aborted) return reject(new Error('iptal'));
-      signal.addEventListener('abort', () => req.destroy(new Error('iptal')), { once: true });
-    }
-  });
-}
-
-async function installEmbeddedPython(signal) {
-  const dest = path.join(beastAppDir(), 'py');
-  fs.mkdirSync(dest, { recursive: true });
-  const zipPath = path.join(os.tmpdir(), 'beast-py-embed.zip');
-  const buf = await httpsGetBuffer(PYTHON_EMBED_URL, 4, signal);
-  fs.writeFileSync(zipPath, buf);
-  const r = await runCommand(
-    `Expand-Archive -LiteralPath "${zipPath}" -DestinationPath "${dest}" -Force`,
-    null,
-    180000
-  );
-  try { fs.unlinkSync(zipPath); } catch {}
-  if (!r.ok || !fs.existsSync(embeddedPythonExe())) {
-    throw new Error('gömülü python kurulamadı: ' + String(r.output || '').slice(0, 200));
-  }
-  /* site-packages yolunu aç — pip ile kurulabilen paketler için ön hazırlık */
-  try {
-    const pth = path.join(dest, 'python312._pth');
-    if (fs.existsSync(pth)) {
-      const cur = fs.readFileSync(pth, 'utf8');
-      if (!/import site/.test(cur)) fs.writeFileSync(pth, cur.replace(/\n*$/, '\nimport site\n'), 'utf8');
-    }
-  } catch {}
-  return { exe: embeddedPythonExe(), prefix: [], source: 'embedded', version: '3.12.8 (gömülü)' };
-}
-
-async function ensurePython({ allowDownload = true, signal } = {}) {
-  if (_pyProbe) {
-    try { fs.accessSync(_pyProbe.exe); return _pyProbe; } catch { _pyProbe = null; }
-  }
-  const emb = embeddedPythonExe();
-  if (fs.existsSync(emb)) {
-    _pyProbe = { exe: emb, prefix: [], source: 'embedded' };
-    return _pyProbe;
-  }
-  const sys = await findSystemPython();
-  if (sys) {
-    _pyProbe = sys;
-    return sys;
-  }
-  if (!allowDownload) throw new Error('python bulunamadı ve otomatik kurulum kapalı');
-  _pyProbe = await installEmbeddedPython(signal);
-  return _pyProbe;
-}
-
-function runPy(exe, scriptPath, cliArgs, cwd, timeoutMs = 120000, signal) {
-  return new Promise((resolve) => {
-    let settled = false;
-    let out = '';
-    let err = '';
-    const startedAt = Date.now();
-    const finish = (code, killed) => {
-      if (settled) return;
-      settled = true;
-      clearTimeout(timer);
-      const text =
-        (out ? out.trim() : '') +
-        (err ? (out ? '\n[stderr] ' : '') + err.trim() : '');
-      resolve({
-        ok: !killed && code === 0,
-        code,
-        ms: Date.now() - startedAt,
-        output: truncateMiddle(text || '(no output)', MAX_CMD_OUTPUT),
-      });
-    };
-
-    let child;
-    try {
-      child = spawn(exe, [scriptPath, ...cliArgs], {
-        cwd: cwd || process.cwd(),
-        windowsHide: true,
-        env: { ...envWithPathPrefix(), PYTHONUTF8: '1', PYTHONIOENCODING: 'utf-8' },
-      });
-    } catch (e) {
-      err += String(e.message);
-      return finish(1, false);
-    }
-
-    const timer = setTimeout(() => {
-      try { spawn('taskkill', ['/pid', String(child.pid), '/T', '/F'], { windowsHide: true }); } catch {}
-      finish(null, true);
-      err += '\n[beast] python timed out';
-    }, timeoutMs);
-
-    if (signal) {
-      if (signal.aborted) return finish(null, true);
-      signal.addEventListener('abort', () => {
-        try { spawn('taskkill', ['/pid', String(child.pid), '/T', '/F'], { windowsHide: true }); } catch {}
-        finish(null, true);
-      }, { once: true });
-    }
-
-    child.stdout.setEncoding('utf8');
-    child.stderr.setEncoding('utf8');
-    child.stdout.on('data', (d) => {
-      out += d;
-      if (out.length > MAX_CMD_OUTPUT * 4) out = out.slice(-MAX_CMD_OUTPUT * 2);
-    });
-    child.stderr.on('data', (d) => {
-      err += d;
-      if (err.length > MAX_CMD_OUTPUT * 4) err = err.slice(-MAX_CMD_OUTPUT * 2);
-    });
-    child.on('error', (e) => { err += String(e.message); finish(1, false); });
-    child.on('close', (code) => finish(code, false));
-  });
-}
-
-/* ---------- web araçları ---------- */
-
-const UA =
-  'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126 Safari/537.36 BeastAgent/0.2';
-
-const PRIVATE_HOST_RE =
-  /^(localhost|.*\.local|127\.|10\.|192\.168\.|169\.254\.|0\.|::1|\[::1\]|f[cd][0-9a-f]{2}:)/i;
-
-function assertPublicHttpUrl(raw) {
-  let u;
-  try {
-    u = new URL(String(raw || ''));
-  } catch {
-    throw new Error('geçersiz URL');
-  }
-  if (u.protocol !== 'http:' && u.protocol !== 'https:') {
-    throw new Error('yalnızca http/https desteklenir');
-  }
-  const host = u.hostname;
-  if (PRIVATE_HOST_RE.test(host)) {
-    throw new Error('yerel/ağ içi adreslere erişim engellendi');
-  }
-  if (/^172\.(1[6-9]|2\d|3[01])\./.test(host)) {
-    throw new Error('yerel/ağ içi adreslere erişim engellendi');
-  }
-  if ((u.username || u.password)) {
-    throw new Error('URL içinde kullanıcı bilgisi desteklenmez');
-  }
-  return u.toString();
-}
-
-const ENTITIES = {
-  amp: '&', lt: '<', gt: '>', quot: '"', apos: "'", nbsp: ' ', mdash: '—', ndash: '–',
-  hellip: '…', rsquo: '\u2019', lsquo: '\u2018', ldquo: '\u201c', rdquo: '\u201d',
-};
-
-function decodeEntities(s) {
-  return String(s || '')
-    .replace(/&#x([0-9a-f]+);/gi, (_m, h) => {
-      try { return String.fromCodePoint(parseInt(h, 16)); } catch { return ''; }
-    })
-    .replace(/&#(\d+);/g, (_m, d) => {
-      try { return String.fromCodePoint(Number(d)); } catch { return ''; }
-    })
-    .replace(/&([a-z]+);/gi, (m, name) => ENTITIES[name.toLowerCase()] ?? m);
-}
-
-function htmlToText(html) {
-  let t = String(html || '');
-  t = t.replace(/<!--[\s\S]*?-->/g, ' ');
-  t = t.replace(/<(script|style|noscript|svg|head)[\s\S]*?<\/\1>/gi, ' ');
-  t = t.replace(/<\/(p|div|li|h[1-6]|tr|section|article|br)>/gi, '\n');
-  t = t.replace(/<br\s*\/?>/gi, '\n');
-  t = t.replace(/<[^>]+>/g, ' ');
-  t = decodeEntities(t);
-  t = t.replace(/[ \t\f\v]+/g, ' ');
-  t = t.replace(/\s*\n\s*/g, '\n').replace(/\n{3,}/g, '\n\n');
-  return t.trim();
-}
-
-/* opencode webfetch.ts tarzı HTML→Markdown (hafif sürüm — Turndown bağımlılığı yok):
-   başlıklar, linkler, kalık/italik, kod blokları, listeler korunur */
-function htmlToMarkdown(html) {
-  let h = String(html || '');
-  h = h.replace(/<script[\s\S]*?<\/script>/gi, '').replace(/<style[\s\S]*?<\/style>/gi, '').replace(/<!--[\s\S]*?-->/g, '');
-  h = h.replace(/<pre[^>]*>\s*<code[^>]*>([\s\S]*?)<\/code>\s*<\/pre>/gi, (_m, c) => '\n```\n' + decodeEntities(c) + '\n```\n');
-  h = h.replace(/<code[^>]*>([\s\S]*?)<\/code>/gi, (_m, c) => '`' + decodeEntities(c) + '`');
-  h = h.replace(/<h([1-6])[^>]*>([\s\S]*?)<\/h\1>/gi, (_m, lvl, c) => '\n' + '#'.repeat(Number(lvl)) + ' ' + decodeEntities(c.replace(/<[^>]+>/g, '')).trim() + '\n');
-  h = h.replace(/<a\s+[^>]*href=["']([^"']+)["'][^>]*>([\s\S]*?)<\/a>/gi, (_m, href, c) => `[${decodeEntities(c.replace(/<[^>]+>/g, '')).trim()}](${href})`);
-  h = h.replace(/<(b|strong)[^>]*>([\s\S]*?)<\/\1>/gi, (_m, _t, c) => `**${decodeEntities(c.replace(/<[^>]+>/g, '')).trim()}**`);
-  h = h.replace(/<(i|em)[^>]*>([\s\S]*?)<\/\1>/gi, (_m, _t, c) => `*${decodeEntities(c.replace(/<[^>]+>/g, '')).trim()}*`);
-  h = h.replace(/<li[^>]*>([\s\S]*?)<\/li>/gi, (_m, c) => `\n- ${decodeEntities(c.replace(/<[^>]+>/g, '')).trim()}`);
-  h = h.replace(/<hr\s*\/?>/gi, '\n---\n');
-  h = h.replace(/<br\s*\/?>/gi, '\n');
-  h = h.replace(/<\/(p|div|section|article|tr|h[1-6])>/gi, '\n');
-  h = h.replace(/<[^>]+>/g, '');
-  h = decodeEntities(h);
-  h = h.replace(/[ \t]+/g, ' ').replace(/\n{3,}/g, '\n\n');
-  return h.trim();
-}
-
-/* DuckDuckGo HTML sonuç ayrıştırıcı — test edilebilir saf fonksiyon */
-function parseDdgResults(html, limit = 8) {
-  const out = [];
-  const re =
-    /<a\s+([^>]*\bresult__a\b[^>]*)>([\s\S]*?)<\/a>([\s\S]*?)(?=<a\s+[^>]*\bresult__a\b|$)/g;
-  let m;
-  while ((m = re.exec(html)) && out.length < limit) {
-    const attrs = m[1] || '';
-    const hrefM = attrs.match(/href="([^"]+)"/) || attrs.match(/href='([^']+)'/);
-    let href = hrefM ? hrefM[1] : '';
-    try {
-      const u = new URL(href, 'https://duckduckgo.com');
-      const uddg = u.searchParams.get('uddg');
-      href = uddg ? decodeURIComponent(uddg) : u.toString();
-    } catch {}
-    const title = decodeEntities(m[2] || '').replace(/<[^>]+>/g, '').trim();
-    const seg = m[3] || '';
-    const snM = seg.match(/<a\s+[^>]*result__snippet[^>]*>([\s\S]*?)<\/a>/);
-    const snippet = snM ? decodeEntities(snM[1]).replace(/<[^>]+>/g, '').trim() : '';
-    if (!title || !/^https?:\/\//i.test(href)) continue;
-    if (out.some((r) => r.url === href)) continue;
-    out.push({ title, url: href, snippet });
-  }
-  return out;
-}
-
-async function fetchWithTimeout(url, opts = {}, timeoutMs = 20000, signal) {
-  const ctrl = new AbortController();
-  const timer = setTimeout(() => ctrl.abort(new Error('zaman aşımı')), timeoutMs);
-  const onAbort = () => ctrl.abort(new Error('iptal'));
-  if (signal) {
-    if (signal.aborted) return Promise.reject(new Error('iptal'));
-    signal.addEventListener('abort', onAbort, { once: true });
-  }
-  try {
-    return await fetch(url, { ...opts, signal: ctrl.signal, redirect: 'follow' });
-  } finally {
-    clearTimeout(timer);
-    if (signal) signal.removeEventListener('abort', onAbort);
-  }
-}
-
-async function webSearch(query, { maxResults = 8, signal } = {}) {
-  const q = String(query || '').trim().slice(0, 400);
-  if (!q) return { ok: false, error: 'boş sorgu' };
-  let res;
-  try {
-    res = await fetchWithTimeout(
-      'https://html.duckduckgo.com/html/',
-      {
-        method: 'POST',
-        headers: {
-          'Content-Type': 'application/x-www-form-urlencoded',
-          'User-Agent': UA,
-          Accept: 'text/html',
-        },
-        body: new URLSearchParams({ q }).toString(),
-      },
-      20000,
-      signal
-    );
-  } catch (e) {
-    /* ağ hatası/rate-limit exception fırlatmasın — zincir devam edebilsin */
-    return { ok: false, error: 'DDG erişilemedi: ' + String((e && e.message) || e) };
-  }
-  if (!res.ok) return { ok: false, error: `DDG HTTP ${res.status}` };
-  const html = await res.text();
-  const results = parseDdgResults(html, Math.min(Math.max(Number(maxResults) || 8, 1), 12));
-  if (!results.length) {
-    return { ok: true, query: q, results, note: 'sonuç yok ya da DDG yanıt biçimi değişti' };
-  }
-  return { ok: true, query: q, count: results.length, results };
-}
-
-const MAX_FETCH_CHARS = 9000;
-
-async function httpFetch(url, { maxChars = MAX_FETCH_CHARS, format = 'text', timeoutMs = 30000, signal } = {}) {
-  const safe = assertPublicHttpUrl(url);
-  const t = Math.min(Math.max(Number(timeoutMs) || 30000, 1000), 120000); // opencode: default 30s, max 120s
-  const res = await fetchWithTimeout(
-    safe,
-    { headers: { 'User-Agent': UA, Accept: 'text/html,text/plain,application/json;q=0.9,*/*;q=0.5' } },
-    t,
-    signal
-  );
-  const ctype = String(res.headers.get('content-type') || '');
-  if (!res.ok) throw new Error(`HTTP ${res.status} ${res.statusText}`);
-  const isTexty =
-    /text\/|json|xml|javascript/i.test(ctype) || ctype === '';
-  if (!isTexty) {
-    return { ok: true, url: safe, status: res.status, contentType: ctype, note: 'ikili/metin olmayan içerik indirilmedi' };
-  }
-  const cap = 400000;
-  let body = await res.text();
-  if (body.length > cap) body = body.slice(0, cap);
-  let content;
-  if (/html/i.test(ctype)) {
-    if (format === 'html') content = body;
-    else if (format === 'markdown') content = htmlToMarkdown(body);
-    else content = htmlToText(body);
-  } else {
-    content = body;
-  }
-  const truncated = content.length > maxChars;
-  return {
-    ok: true,
-    url: safe,
-    status: res.status,
-    contentType: ctype,
-    format,
-    truncated,
-    content: content.slice(0, Math.max(1000, Math.min(Number(maxChars) || MAX_FETCH_CHARS, 50000))),
-  };
-}
-
-/* ================= opencode edit.ts BİREBİR PORT =================
-   Kaynak: opencode-dev/packages/opencode/src/tool/edit.ts
-   9 aşamalı replacer zinciri: Simple → LineTrimmed → BlockAnchor →
-   WhitespaceNormalized → IndentationFlexible → EscapeNormalized →
-   TrimmedBoundary → ContextAware → MultiOccurrence.
-   Modelin old_string'i ufak girinti/boşluk farkıyla tutturamadığında zincir
-   akıllı eşleşme bulur — "dosyayı baştan oku" döngüsü kökten kırılır. */
-
-function normalizeLineEndings(text) {
-  return text.replaceAll('\r\n', '\n');
-}
-function detectLineEnding(text) {
-  return text.includes('\r\n') ? '\r\n' : '\n';
-}
-function convertToLineEnding(text, ending) {
-  if (ending === '\n') return text;
-  return text.replaceAll('\n', '\r\n');
-}
-
-/* blok-anchor fallback benzerlik eşikleri (edit.ts:220-221) */
-const SINGLE_CANDIDATE_SIMILARITY_THRESHOLD = 0.65;
-const MULTIPLE_CANDIDATES_SIMILARITY_THRESHOLD = 0.65;
-
-function levenshtein(a, b) {
-  if (a === '' || b === '') return Math.max(a.length, b.length);
-  const matrix = Array.from({ length: a.length + 1 }, (_, i) =>
-    Array.from({ length: b.length + 1 }, (_, j) => (i === 0 ? j : j === 0 ? i : 0))
-  );
-  for (let i = 1; i <= a.length; i++) {
-    for (let j = 1; j <= b.length; j++) {
-      const cost = a[i - 1] === b[j - 1] ? 0 : 1;
-      matrix[i][j] = Math.min(matrix[i - 1][j] + 1, matrix[i][j - 1] + 1, matrix[i - 1][j - 1] + cost);
-    }
-  }
-  return matrix[a.length][b.length];
-}
-
-const SimpleReplacer = function* (_content, find) {
-  yield find;
-};
-
-const LineTrimmedReplacer = function* (content, find) {
-  const originalLines = content.split('\n');
-  const searchLines = find.split('\n');
-  if (searchLines[searchLines.length - 1] === '') searchLines.pop();
-  for (let i = 0; i <= originalLines.length - searchLines.length; i++) {
-    let matches = true;
-    for (let j = 0; j < searchLines.length; j++) {
-      if (originalLines[i + j].trim() !== searchLines[j].trim()) {
-        matches = false;
-        break;
-      }
-    }
-    if (matches) {
-      let matchStartIndex = 0;
-      for (let k = 0; k < i; k++) matchStartIndex += originalLines[k].length + 1;
-      let matchEndIndex = matchStartIndex;
-      for (let k = 0; k < searchLines.length; k++) {
-        matchEndIndex += originalLines[i + k].length;
-        if (k < searchLines.length - 1) matchEndIndex += 1;
-      }
-      yield content.substring(matchStartIndex, matchEndIndex);
-    }
-  }
-};
-
-const BlockAnchorReplacer = function* (content, find) {
-  const originalLines = content.split('\n');
-  const searchLines = find.split('\n');
-  if (searchLines.length < 3) return;
-  if (searchLines[searchLines.length - 1] === '') searchLines.pop();
-  const firstLineSearch = searchLines[0].trim();
-  const lastLineSearch = searchLines[searchLines.length - 1].trim();
-  const searchBlockSize = searchLines.length;
-  const maxLineDelta = Math.max(1, Math.floor(searchBlockSize * 0.25));
-  const candidates = [];
-  for (let i = 0; i < originalLines.length; i++) {
-    if (originalLines[i].trim() !== firstLineSearch) continue;
-    for (let j = i + 2; j < originalLines.length; j++) {
-      if (originalLines[j].trim() === lastLineSearch) {
-        const actualBlockSize = j - i + 1;
-        if (Math.abs(actualBlockSize - searchBlockSize) <= maxLineDelta) {
-          candidates.push({ startLine: i, endLine: j });
-        }
-        break;
-      }
-    }
-  }
-  if (candidates.length === 0) return;
-  if (candidates.length === 1) {
-    const { startLine, endLine } = candidates[0];
-    const actualBlockSize = endLine - startLine + 1;
-    let similarity = 0;
-    const linesToCheck = Math.min(searchBlockSize - 2, actualBlockSize - 2);
-    if (linesToCheck > 0) {
-      for (let j = 1; j < searchBlockSize - 1 && j < actualBlockSize - 1; j++) {
-        const originalLine = originalLines[startLine + j].trim();
-        const searchLine = searchLines[j].trim();
-        const maxLen = Math.max(originalLine.length, searchLine.length);
-        if (maxLen === 0) continue;
-        const distance = levenshtein(originalLine, searchLine);
-        similarity += (1 - distance / maxLen) / linesToCheck;
-        if (similarity >= SINGLE_CANDIDATE_SIMILARITY_THRESHOLD) break;
-      }
-    } else {
-      similarity = 1.0;
-    }
-    if (similarity >= SINGLE_CANDIDATE_SIMILARITY_THRESHOLD) {
-      let matchStartIndex = 0;
-      for (let k = 0; k < startLine; k++) matchStartIndex += originalLines[k].length + 1;
-      let matchEndIndex = matchStartIndex;
-      for (let k = startLine; k <= endLine; k++) {
-        matchEndIndex += originalLines[k].length;
-        if (k < endLine) matchEndIndex += 1;
-      }
-      yield content.substring(matchStartIndex, matchEndIndex);
-    }
-    return;
-  }
-  let bestMatch = null;
-  let maxSimilarity = -1;
-  for (const candidate of candidates) {
-    const { startLine, endLine } = candidate;
-    const actualBlockSize = endLine - startLine + 1;
-    let similarity = 0;
-    const linesToCheck = Math.min(searchBlockSize - 2, actualBlockSize - 2);
-    if (linesToCheck > 0) {
-      for (let j = 1; j < searchBlockSize - 1 && j < actualBlockSize - 1; j++) {
-        const originalLine = originalLines[startLine + j].trim();
-        const searchLine = searchLines[j].trim();
-        const maxLen = Math.max(originalLine.length, searchLine.length);
-        if (maxLen === 0) continue;
-        const distance = levenshtein(originalLine, searchLine);
-        similarity += 1 - distance / maxLen;
-      }
-      similarity /= linesToCheck;
-    } else {
-      similarity = 1.0;
-    }
-    if (similarity > maxSimilarity) {
-      maxSimilarity = similarity;
-      bestMatch = candidate;
-    }
-  }
-  if (maxSimilarity >= MULTIPLE_CANDIDATES_SIMILARITY_THRESHOLD && bestMatch) {
-    const { startLine, endLine } = bestMatch;
-    let matchStartIndex = 0;
-    for (let k = 0; k < startLine; k++) matchStartIndex += originalLines[k].length + 1;
-    let matchEndIndex = matchStartIndex;
-    for (let k = startLine; k <= endLine; k++) {
-      matchEndIndex += originalLines[k].length;
-      if (k < endLine) matchEndIndex += 1;
-    }
-    yield content.substring(matchStartIndex, matchEndIndex);
-  }
-};
-
-const WhitespaceNormalizedReplacer = function* (content, find) {
-  const normalizeWhitespace = (text) => text.replace(/\s+/g, ' ').trim();
-  const normalizedFind = normalizeWhitespace(find);
-  const lines = content.split('\n');
-  for (let i = 0; i < lines.length; i++) {
-    const line = lines[i];
-    if (normalizeWhitespace(line) === normalizedFind) {
-      yield line;
-    } else {
-      const normalizedLine = normalizeWhitespace(line);
-      if (normalizedLine.includes(normalizedFind)) {
-        const words = find.trim().split(/\s+/);
-        if (words.length > 0) {
-          const pattern = words.map((word) => word.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')).join('\\s+');
-          try {
-            const regex = new RegExp(pattern);
-            const match = line.match(regex);
-            if (match) yield match[0];
-          } catch {}
-        }
-      }
-    }
-  }
-  const findLines = find.split('\n');
-  if (findLines.length > 1) {
-    for (let i = 0; i <= lines.length - findLines.length; i++) {
-      const block = lines.slice(i, i + findLines.length);
-      if (normalizeWhitespace(block.join('\n')) === normalizedFind) {
-        yield block.join('\n');
-      }
-    }
-  }
-};
-
-const IndentationFlexibleReplacer = function* (content, find) {
-  const removeIndentation = (text) => {
-    const lines = text.split('\n');
-    const nonEmptyLines = lines.filter((line) => line.trim().length > 0);
-    if (nonEmptyLines.length === 0) return text;
-    const minIndent = Math.min(
-      ...nonEmptyLines.map((line) => {
-        const match = line.match(/^(\s*)/);
-        return match ? match[1].length : 0;
-      })
-    );
-    return lines.map((line) => (line.trim().length === 0 ? line : line.slice(minIndent))).join('\n');
-  };
-  const normalizedFind = removeIndentation(find);
-  const contentLines = content.split('\n');
-  const findLines = find.split('\n');
-  for (let i = 0; i <= contentLines.length - findLines.length; i++) {
-    const block = contentLines.slice(i, i + findLines.length).join('\n');
-    if (removeIndentation(block) === normalizedFind) yield block;
-  }
-};
-
-const EscapeNormalizedReplacer = function* (content, find) {
-  const unescapeString = (str) =>
-    str.replace(/\\(n|t|r|'|"|`|\\|\n|\$)/g, (match, capturedChar) => {
-      switch (capturedChar) {
-        case 'n': return '\n';
-        case 't': return '\t';
-        case 'r': return '\r';
-        case "'": return "'";
-        case '"': return '"';
-        case '`': return '`';
-        case '\\': return '\\';
-        case '\n': return '\n';
-        case '$': return '$';
-        default: return match;
-      }
-    });
-  const unescapedFind = unescapeString(find);
-  if (content.includes(unescapedFind)) yield unescapedFind;
-  const lines = content.split('\n');
-  const findLines = unescapedFind.split('\n');
-  for (let i = 0; i <= lines.length - findLines.length; i++) {
-    const block = lines.slice(i, i + findLines.length).join('\n');
-    if (unescapeString(block) === unescapedFind) yield block;
-  }
-};
-
-const MultiOccurrenceReplacer = function* (content, find) {
-  let startIndex = 0;
-  while (true) {
-    const index = content.indexOf(find, startIndex);
-    if (index === -1) break;
-    yield find;
-    startIndex = index + find.length;
-  }
-};
-
-const TrimmedBoundaryReplacer = function* (content, find) {
-  const trimmedFind = find.trim();
-  if (trimmedFind === find) return;
-  if (content.includes(trimmedFind)) yield trimmedFind;
-  const lines = content.split('\n');
-  const findLines = find.split('\n');
-  for (let i = 0; i <= lines.length - findLines.length; i++) {
-    const block = lines.slice(i, i + findLines.length).join('\n');
-    if (block.trim() === trimmedFind) yield block;
-  }
-};
-
-const ContextAwareReplacer = function* (content, find) {
-  const findLines = find.split('\n');
-  if (findLines.length < 3) return;
-  if (findLines[findLines.length - 1] === '') findLines.pop();
-  const contentLines = content.split('\n');
-  const firstLine = findLines[0].trim();
-  const lastLine = findLines[findLines.length - 1].trim();
-  for (let i = 0; i < contentLines.length; i++) {
-    if (contentLines[i].trim() !== firstLine) continue;
-    for (let j = i + 2; j < contentLines.length; j++) {
-      if (contentLines[j].trim() === lastLine) {
-        const blockLines = contentLines.slice(i, j + 1);
-        const block = blockLines.join('\n');
-        if (blockLines.length === findLines.length) {
-          let matchingLines = 0;
-          let totalNonEmptyLines = 0;
-          for (let k = 1; k < blockLines.length - 1; k++) {
-            const blockLine = blockLines[k].trim();
-            const findLine = findLines[k].trim();
-            if (blockLine.length > 0 || findLine.length > 0) {
-              totalNonEmptyLines++;
-              if (blockLine === findLine) matchingLines++;
-            }
-          }
-          if (totalNonEmptyLines === 0 || matchingLines / totalNonEmptyLines >= 0.5) {
-            yield block;
-            break;
-          }
-        }
-        break;
-      }
-    }
-  }
-};
-
-function isDisproportionateMatch(search, oldString) {
-  const oldLines = oldString.split('\n').length;
-  const searchLines = search.split('\n').length;
-  if (searchLines >= Math.max(oldLines + 3, oldLines * 2)) return true;
-  if (oldLines === 1) return false;
-  return search.trim().length > Math.max(oldString.trim().length + 500, oldString.trim().length * 4);
-}
-
-/* edit.ts:682-729 replace() — zinciri sırayla dener; tek eşleşme şart,
-   replaceAll'de tümünü değiştirir; hata mesajları BİREBİR opencode */
-function ocReplace(content, oldString, newString, replaceAll = false) {
-  if (oldString === newString) {
-    throw new Error('No changes to apply: oldString and newString are identical.');
-  }
-  if (oldString === '') {
-    throw new Error(
-      'oldString cannot be empty when editing an existing file. Provide the exact text to replace, or use write_file for an intentional full-file replacement.'
-    );
-  }
-  let notFound = true;
-  for (const replacer of [
-    SimpleReplacer,
-    LineTrimmedReplacer,
-    BlockAnchorReplacer,
-    WhitespaceNormalizedReplacer,
-    IndentationFlexibleReplacer,
-    EscapeNormalizedReplacer,
-    TrimmedBoundaryReplacer,
-    ContextAwareReplacer,
-    MultiOccurrenceReplacer,
-  ]) {
-    for (const search of replacer(content, oldString)) {
-      const index = content.indexOf(search);
-      if (index === -1) continue;
-      notFound = false;
-      if (isDisproportionateMatch(search, oldString)) {
-        throw new Error(
-          'Refusing replacement because the matched span is much larger than oldString. Re-read the file and provide the full exact oldString for the intended replacement.'
-        );
-      }
-      if (replaceAll) {
-        return { result: content.replaceAll(search, newString), replacements: content.split(search).length - 1 };
-      }
-      const lastIndex = content.lastIndexOf(search);
-      if (index !== lastIndex) continue;
-      return { result: content.substring(0, index) + newString + content.substring(index + search.length), replacements: 1 };
-    }
-  }
-  if (notFound) {
-    throw new Error(
-      'Could not find oldString in the file. It must match exactly, including whitespace, indentation, and line endings.'
-    );
-  }
-  throw new Error('Found multiple matches for oldString. Provide more surrounding context to make the match unique.');
-}
-
-/* ---------- opencode diff portu (npm "diff" paketinin diffLines karşılığı) ----------
-   LCS tabanlı satır diff'i: additions/deletions sayımı (edit.ts:175-180) ve
-   UI split-diff görünümü için kırpılmış bölge (trimDiff mantığı). */
-function diffLineCounts(aText, bText) {
-  /* bos metin = 0 satir (opencode npm diffLines davranisi) */
-  const a = aText ? String(aText).split('\n') : [];
-  const b = bText ? String(bText).split('\n') : [];
-  let p = 0;
-  while (p < a.length && p < b.length && a[p] === b[p]) p++;
-  let ea = a.length - 1;
-  let eb = b.length - 1;
-  while (ea >= p && eb >= p && a[ea] === b[eb]) { ea--; eb--; }
-  const midA = a.slice(p, ea + 1);
-  const midB = b.slice(p, eb + 1);
-  const n = midA.length;
-  const m = midB.length;
-  if (!n && !m) return { additions: 0, deletions: 0 };
-  if (n * m > 400000 || n > 1500 || m > 1500) return { additions: m, deletions: n };
-  const w = m + 1;
-  const dp = new Int32Array((n + 1) * w);
-  for (let i = n - 1; i >= 0; i--) {
-    for (let j = m - 1; j >= 0; j--) {
-      dp[i * w + j] =
-        midA[i] === midB[j]
-          ? dp[(i + 1) * w + j + 1] + 1
-          : Math.max(dp[(i + 1) * w + j], dp[i * w + j + 1]);
-    }
-  }
-  const matched = dp[0];
-  return { additions: m - matched, deletions: n - matched };
-}
-
-const DIFF_REGION_CTX = 3; /* değişiklik etrafında kaç bağlam satırı gösterilir */
-const DIFF_REGION_CAP = 3500; /* UI'ye giden bölge başına karakter tavanı */
-
-function capDiffText(s) {
-  const t = String(s || '');
-  return t.length > DIFF_REGION_CAP ? t.slice(0, DIFF_REGION_CAP) + '\n…' : t;
-}
-
-/* Değişen bölgeyi ±DIFF_REGION_CTX bağlam satırıyla kırpar (opencode trimDiff
-   mantığı — tüm dosya yerine yalnız değişen kısım UI'ye gider). */
-function diffRegion(before, after) {
-  const a = String(before || '').split('\n');
-  const b = String(after || '').split('\n');
-  let p = 0;
-  while (p < a.length && p < b.length && a[p] === b[p]) p++;
-  let ea = a.length - 1;
-  let eb = b.length - 1;
-  while (ea >= p && eb >= p && a[ea] === b[eb]) { ea--; eb--; }
-  const start = Math.max(0, p - DIFF_REGION_CTX);
-  const beforeRegion = a.slice(start, Math.min(a.length, ea + 1 + DIFF_REGION_CTX)).join('\n');
-  const afterRegion = b.slice(start, Math.min(b.length, eb + 1 + DIFF_REGION_CTX)).join('\n');
-  return { before: capDiffText(beforeRegion), after: capDiffText(afterRegion), startLine: start + 1 };
-}
-
-/* ---------- read_file disk-cache (opencode file-state portu) ----------
-   mtime+size değişmemişse dosya yeniden diskten OKUNMAZ — cache'ten döner.
-   edit_file/write_file kendi yazdıklarından sonra cache'i düşürür; böylece
-   sonraki okuma daima güncel içerik verir. */
-const _readCache = new Map(); /* abs → { mtimeMs, size, raw } */
-
-function readCacheGet(abs, st) {
-  const hit = _readCache.get(abs);
-  if (hit && hit.mtimeMs === st.mtimeMs && hit.size === st.size) return hit.raw;
-  const raw = fs.readFileSync(abs, 'utf8');
-  if (_readCache.size > 200) _readCache.delete(_readCache.keys().next().value);
-  _readCache.set(abs, { mtimeMs: st.mtimeMs, size: st.size, raw });
-  return raw;
-}
-
-function readCacheDrop(abs) {
-  _readCache.delete(abs);
-}
-
-const definitions = [
-  {
-    type: 'function',
-    function: {
-      name: 'run_command',
-      description:
-        'Executes a given PowerShell command on the user\'s Windows machine in the workspace directory and returns combined stdout/stderr. Use this tool for terminal operations like builds, git, npm installs, docker, etc. DO NOT use it for file operations (reading, writing, editing, searching, finding files) - use the specialized tools for this instead: read_file, write_file, edit_file, grep, glob.',
-      parameters: {
-        type: 'object',
-        properties: {
-          command: { type: 'string', description: 'PowerShell command line to execute' },
-          timeout_ms: { type: 'number', description: 'Optional timeout in ms (default 120000)' },
-        },
-        required: ['command'],
-      },
-    },
-  },
-  {
-    type: 'function',
-    function: {
-      name: 'read_file',
-      /* opencode read.txt BİREBİR port (parametre adı path olarak kaldı) */
-      description:
-        'Read a file or directory from the local filesystem. If the path does not exist, an error is returned.\n\n' +
-        'Usage:\n' +
-        '- By default, this tool returns up to 2000 lines from the start of the file.\n' +
-        '- The offset parameter is the line number to start reading from (1-indexed).\n' +
-        '- To read later sections, call this tool again with a larger offset — NEVER re-read from the start of the file.\n' +
-        '- Use the grep tool to find specific content in large files or files with long lines.\n' +
-        '- If you are unsure of the correct file path, use the glob tool to look up filenames by glob pattern.\n' +
-        '- Contents are returned with each line prefixed by its line number as `<line>: <content>`. For example, if a file has contents "foo\\n", you will receive "1: foo\\n".\n' +
-        '- Any line longer than 2000 characters is truncated.\n' +
-        '- Call this tool in parallel when you know there are multiple files you want to read.\n' +
-        '- Avoid tiny repeated slices (30 line chunks). If you need more context, read a larger window.\n' +
-        '- File contents you read STAY in your context for the WHOLE session. Do NOT re-read a file you already read: chain edits from your previous read plus your own edit_file/write_file results. Re-reading the same file (same or overlapping range) wastes turns and is forbidden unless the file actually changed on disk outside your own edits.',
-      parameters: {
-        type: 'object',
-        properties: {
-          path: { type: 'string', description: 'The path to the file to read' },
-          offset: { type: 'number', description: 'The line number to start reading from (1-indexed)' },
-          limit: { type: 'number', description: 'The maximum number of lines to read (defaults to 2000)' },
-        },
-        required: ['path'],
-      },
-    },
-  },
-  {
-    type: 'function',
-    function: {
-      name: 'write_file',
-      /* opencode write.txt BİREBİR port */
-      description:
-        'Writes a file to the local filesystem.\n\n' +
-        'Usage:\n' +
-        '- This tool will overwrite the existing file if there is one at the provided path.\n' +
-        '- If this is an existing file, you MUST use the read_file tool first to read the file\'s contents. This tool will fail if you did not read the file first.\n' +
-        '- ALWAYS prefer editing existing files in the codebase with edit_file. NEVER write new files unless explicitly required.\n' +
-        '- The result includes additions/deletions counts — the change is APPLIED to disk immediately; do NOT read the file again to verify.',
-      parameters: {
-        type: 'object',
-        properties: {
-          path: { type: 'string', description: 'The path to the file to write' },
-          content: { type: 'string', description: 'The content to write to the file' },
-        },
-        required: ['path', 'content'],
-      },
-    },
-  },
-  {
-    type: 'function',
-    function: {
-      name: 'edit_file',
-      /* opencode edit.txt BİREBİR port (parametre adları snake_case kaldı) */
-      description:
-        'Performs exact string replacements in files.\n\n' +
-        'Usage:\n' +
-        '- You must use your read_file tool at least once in the conversation before editing. This tool will error if you attempt an edit without reading the file.\n' +
-        '- When editing text from read_file output, ensure you preserve the exact indentation (tabs/spaces) as it appears AFTER the line number prefix. The line number prefix format is: line number + colon + space (e.g., `1: `). Everything after that space is the actual file content to match. Never include any part of the line number prefix in the old_string or new_string.\n' +
-        '- ALWAYS prefer editing existing files in the codebase. NEVER write new files unless explicitly required.\n' +
-        '- The edit will FAIL if `old_string` is not found in the file with an error "oldString not found in content".\n' +
-        '- The edit will FAIL if `old_string` is found multiple times in the file with an error "Found multiple matches for oldString. Provide more surrounding lines in old_string to identify the correct match." Either provide a larger string with more surrounding context to make it unique or use `replace_all` to change every instance of `old_string`.\n' +
-        '- Use `replace_all` for replacing and renaming strings across the file. This parameter is useful if you want to rename a variable for instance.\n' +
-        '- The result includes additions/deletions counts — the change is APPLIED to disk immediately; do NOT read the file again to verify.\n' +
-        '- Chain follow-up edits from what you already read plus your own previous edits — do NOT re-read the file between edits. Multiple edit_file calls to different files can be issued in PARALLEL in one turn.',
-      parameters: {
-        type: 'object',
-        properties: {
-          path: { type: 'string', description: 'The path to the file to modify' },
-          old_string: { type: 'string', description: 'The text to replace' },
-          new_string: { type: 'string', description: 'The text to replace it with (must be different from old_string)' },
-          replace_all: { type: 'boolean', description: 'Replace all occurrences of old_string (default false)' },
-        },
-        required: ['path', 'old_string', 'new_string'],
-      },
-    },
-  },
-  {
-    type: 'function',
-    function: {
-      name: 'grep',
-      description:
-        'Fast content search tool that works with any codebase size. Searches file contents using regular expressions (case-SENSITIVE by default — pass case_insensitive:true to ignore case); supports full regex syntax (eg. "log.*Error", "function\\s+\\w+"). Filter files by pattern with include (eg. "*.js", "*.{ts,tsx}"). Returns grouped matches as `<path>:` + `  Line N: text`, capped at 100 matches. node_modules/.git and similar dirs are skipped. Use this to find where functions/symbols/errors live before editing.',
-      parameters: {
-        type: 'object',
-        properties: {
-          pattern: { type: 'string', description: 'Regular expression' },
-          path: { type: 'string', description: 'File or directory to search; defaults to workspace root' },
-          include: { type: 'string', description: 'Glob filter like "*.js" or "*.{ts,tsx}"' },
-          case_insensitive: { type: 'boolean', description: 'Ignore case (default false — search is case-sensitive)' },
-        },
-        required: ['pattern'],
-      },
-    },
-  },
-  {
-    type: 'function',
-    function: {
-      name: 'glob',
-      description:
-        'Fast file pattern matching tool that works with any codebase size. Supports glob patterns like "**/*.js" or "src/**/*.ts". Returns matching file paths (max 100; more specific pattern/path if truncated). Use when you need to find files by name patterns; batch multiple speculative searches in one turn.',
-      parameters: {
-        type: 'object',
-        properties: {
-          pattern: { type: 'string', description: 'Glob pattern, e.g. "src/**/*.ts"' },
-          path: { type: 'string', description: 'Directory to search; defaults to workspace root' },
-        },
-        required: ['pattern'],
-      },
-    },
-  },
-  {
-    type: 'function',
-    function: {
-      name: 'list_dir',
-      description: 'List entries of a directory (localeCompare sorted, directories suffixed with `/`, max 500 entries) with sizes and types.',
-      parameters: {
-        type: 'object',
-        properties: {
-          path: { type: 'string', description: 'Defaults to workspace root' },
-        },
-        required: [],
-      },
-    },
-  },
-  {
-    type: 'function',
-    function: {
-      name: 'webfetch',
-      description:
-        '- Fetches content from a specified URL\n' +
-        '- Takes a URL and optional format as input\n' +
-        '- Fetches the URL content, converts to requested format (markdown by default)\n' +
-        '- Returns the content in the specified format\n' +
-        '- Use this tool when you need to retrieve and analyze web content\n' +
-        '- IMPORTANT: if another tool is present that offers better web fetching capabilities, is more targeted to the task, or has fewer restrictions, prefer using that tool instead of this one.\n' +
-        '- The URL must be a fully-formed valid URL\n' +
-        '- HTTP URLs will be automatically upgraded to HTTPS\n' +
-        '- Format options: "text" (default), "markdown", or "html"\n' +
-        '- Local/private network addresses are blocked; timeout default 30s (max 120s)',
-      parameters: {
-        type: 'object',
-        properties: {
-          url: { type: 'string', format: 'uri', description: 'The URL to fetch content from' },
-          format: { type: 'string', enum: ['text', 'markdown', 'html'], description: 'The format to return the content in (defaults to text)' },
-          timeout: { type: 'number', description: 'Optional timeout in seconds (max 120)' },
-          max_chars: { type: 'number', description: 'Output character cap (default 9000, max 50000)' },
-        },
-        required: ['url'],
-      },
-    },
-  },
-  {
-    type: 'function',
-    function: {
-      name: 'web_search',
-      description:
-        'FAST web search with an automatic chain (order is configurable in Ayarlar → Web Arama): built-in browser (real Chromium searching GOOGLE directly with AI Mode — no bot protection; the response may include an `ai` field holding Google\'s own AI answer), Obscura stealth headless browser (anti-detect; searches DuckDuckGo, installed automatically and ACTIVE by default), TinyFish API (only if a key is configured), then Python multi-engine (ddgs / DuckDuckGo+Bing+Mojeek). If the browser hits CAPTCHA/unusual traffic it is skipped for 10 minutes and the next engine takes over. Returns {ai?, results[{title,url,snippet}]}. Use when fresh or external info is needed; skip for things you already know.',
-      parameters: {
-        type: 'object',
-        properties: {
-          query: { type: 'string' },
-          max_results: { type: 'number', description: '1-12, default 8' },
-        },
-        required: ['query'],
-      },
-    },
-  },
-  {
-    type: 'function',
-    function: {
-      name: 'http_fetch',
-      description:
-        'Fetch one URL and return its text content (HTML converted to plain text, JSON as-is). Local/private network addresses are blocked.',
-      parameters: {
-        type: 'object',
-        properties: {
-          url: { type: 'string', format: 'uri' },
-          max_chars: { type: 'number', description: 'default 9000' },
-        },
-        required: ['url'],
-      },
-    },
-  },
-  {
-    type: 'function',
-    function: {
-      name: 'deep_search',
-      description:
-        'AGENTIC DEEP RESEARCH — use when web_search is not enough or the answer was NOT found on the first try. Runs 1-4 query variants IN PARALLEL (rephrase, synonyms, Turkish + English spellings of the same question), merges + dedupes results, then AUTOMATICALLY opens and reads the top pages with a HIDDEN real-Chromium browser (JS/SPA pages work; the visible panel is NOT touched) and returns full-text excerpts. Ideal for: multi-angle questions (price comparison, reviews, specs, "find everything about X"), Turkish queries that miss results, pages that need JS rendering. Returns {queries, results[{title,url,snippet}], pages[{url,title,content}]}.',
-      parameters: {
-        type: 'object',
-        properties: {
-          queries: {
-            type: 'array',
-            maxItems: 4,
-            items: { type: 'string' },
-            description: '1-4 query variants; e.g. ["iphone 16 fiyat", "iphone 16 price turkey", "iphone 16 technosa"]',
-          },
-          max_results: { type: 'number', description: 'merged result cap, default 16' },
-          read_top: { type: 'number', description: 'how many top results to auto-read fully (0-6, default 3); 0 = results only' },
-        },
-        required: ['queries'],
-      },
-    },
-  },
-  {
-    type: 'function',
-    function: {
-      name: 'python_run',
-      description:
-        "Run Python (system interpreter, or Beast auto-installs a portable embedded runtime on first use). Either pass inline `code`, or `script` = filename inside the Beast scripts library (%APPDATA%\\beast\\scripts — e.g. news.py haber toplayıcı) or an absolute path. Use for scraping, RSS/haber toplama, veri analizi, regex/parsing işleri. stdout+stderr returned; scripts should print results.",
-      parameters: {
-        type: 'object',
-        properties: {
-          code: { type: 'string', description: 'Inline Python source (use this OR script)' },
-          script: { type: 'string', description: 'Script name in beast scripts dir or absolute path' },
-          args: {
-            type: 'array',
-            items: { type: 'string' },
-            description: 'CLI arguments passed to the script, e.g. ["--limit","5","--json"]',
-          },
-          timeout_ms: { type: 'number', description: 'default 120000' },
-        },
-      },
-    },
-  },
-];
-
-async function exec(name, args, ctx) {
-  const cwd = ctx.cwd;
-  try {
-    switch (name) {
-      case 'run_command': {
-        /* opencode shell.ts kuralları: default 120s, negatif timeout reddi */
-        const tRaw = Number(args.timeout_ms);
-        if (Number.isFinite(tRaw) && args.timeout_ms != null && tRaw <= 0) {
-          return JSON.stringify({ ok: false, error: `Invalid timeout value: ${tRaw}. Timeout must be a positive number.` });
-        }
-        const t = Number.isFinite(tRaw) && tRaw > 0 ? tRaw : 120000;
-        const shell = String(args.shell || 'powershell').toLowerCase();
-        const r =
-          shell === 'bash' || shell === 'sh'
-            ? await runBashCommand(String(args.command || ''), cwd, t, ctx.signal)
-            : await runShellCommand(String(args.command || ''), cwd, t, ctx.signal);
-        const b = boundToolOutput((r && r.output) || '');
-        return JSON.stringify({
-          ok: !!(r && r.ok),
-          code: r && r.code,
-          output: b.text || '(no output)',
-          ...(b.outputFile ? { outputFile: b.outputFile } : {}),
-        });
-      }
-      case 'edit_file': {
-        /* opencode edit.ts execute BİREBİR port — replacer zinciri + satır sonu
-           normalizasyonu + diff metadata (additions/deletions + UI diffView) */
-        const filePath = safeResolve(String(args.path || args.filePath || ''), cwd);
-        const oldS = String(args.old_string ?? args.oldString ?? '');
-        const newS = String(args.new_string ?? args.newString ?? '');
-        try {
-          if (oldS === newS) {
-            throw new Error('No changes to apply: oldString and newString are identical.');
-          }
-          /* boş oldString = YENİ dosya oluşturma yolu (edit.ts:90-121) */
-          if (oldS === '') {
-            if (fs.existsSync(filePath)) {
-              throw new Error(
-                'oldString cannot be empty when editing an existing file. Provide the exact text to replace, or use write_file for an intentional full-file replacement.'
-              );
-            }
-            fs.mkdirSync(path.dirname(filePath), { recursive: true });
-            fs.writeFileSync(filePath, newS, 'utf8');
-            readCacheDrop(filePath);
-            const counts = diffLineCounts('', newS);
-            const out = { ok: true, path: filePath, note: 'Edit applied successfully.', replacements: 1, ...counts };
-            if (ctx.wantDiff) out.diffView = { path: filePath, before: '', after: capDiffText(newS), startLine: 1, ...counts };
-            return JSON.stringify(out);
-          }
-          if (!fs.existsSync(filePath)) throw new Error(`File ${filePath} not found`);
-          if (fs.statSync(filePath).isDirectory()) throw new Error(`Path is a directory, not a file: ${filePath}`);
-          const contentOld = fs.readFileSync(filePath, 'utf8');
-          /* satır sonu stili dosyadan alınır, old/new buna çevrilir (edit.ts:129-131) */
-          const ending = detectLineEnding(contentOld);
-          const oldN = convertToLineEnding(normalizeLineEndings(oldS), ending);
-          const newN = convertToLineEnding(normalizeLineEndings(newS), ending);
-          const rep = ocReplace(contentOld, oldN, newN, !!(args.replace_all || args.replaceAll));
-          const contentNew = rep.result;
-          fs.writeFileSync(filePath, contentNew, 'utf8');
-          readCacheDrop(filePath); /* sonraki read_file taze içerik okusun */
-          const counts = diffLineCounts(contentOld, contentNew);
-          const out = {
-            ok: true,
-            path: filePath,
-            note: 'Edit applied successfully.',
-            replacements: rep.replacements,
-            ...counts,
-          };
-          if (ctx.wantDiff) {
-            out.diffView = { path: filePath, ...diffRegion(contentOld, contentNew), ...counts };
-          }
-          return JSON.stringify(out);
-        } catch (e) {
-          return JSON.stringify({ ok: false, error: String((e && e.message) || e) });
-        }
-      }
-      case 'read_file': {
-        /* opencode read.ts BİREBİR port: fuzzy not-found, binary tespiti,
-           dizin okuma, `N: içerik` formatı, 2000 satır / 2000 karakter /
-           50KB tavanları + opencode devam footer'ları */
-        const abs = safeResolve(String(args.path || ''), cwd);
-        if (!fs.existsSync(abs)) {
-          const sibs = fuzzySiblings(abs);
-          const hint = sibs.length ? '\nDid you mean one of these?\n' + sibs.map((s) => `- ${path.join(path.dirname(abs), s)}`).join('\n') : '';
-          return JSON.stringify({ ok: false, error: `File not found: ${abs}${hint}` });
-        }
-        const st = fs.statSync(abs);
-        /* DİZİN OKUMA (opencode read.ts dizin modu): localeCompare sıralı,
-           klasörler `/` ile, offset/limit sayfalı */
-        if (st.isDirectory()) {
-          let names = fs
-            .readdirSync(abs)
-            .sort((a, b) => a.localeCompare(b))
-            .map((name) => {
-              let isDir = false;
-              try { isDir = fs.statSync(path.join(abs, name)).isDirectory(); } catch {}
-              return isDir ? name + '/' : name;
-            });
-          const total = names.length;
-          const offset = Math.max(1, Math.floor(Number(args.offset) || 1));
-          const limit = Math.min(2000, Math.max(1, Math.floor(Number(args.limit) || 2000)));
-          names = names.slice(offset - 1, offset - 1 + limit);
-          const note =
-            names.length < total
-              ? `(Showing ${names.length} of ${total} entries. Use offset parameter to paginate.)`
-              : `(End of directory - total ${total} entries)`;
-          return JSON.stringify({
-            ok: true,
-            path: abs,
-            type: 'directory',
-            totalEntries: total,
-            offset,
-            truncated: offset - 1 + names.length < total,
-            note,
-            content: names.join('\n'),
-          });
-        }
-        if (st.size > MAX_FILE_CHARS * 2) {
-          return JSON.stringify({ ok: false, error: `file too large (${st.size} bytes)` });
-        }
-        /* binary tespiti uzantıdan — PDF'e dokunma (Beast'in pdf hattı var) */
-        if (BINARY_EXT_RE.test(abs)) {
-          return JSON.stringify({ ok: false, error: `Cannot read binary file: ${abs}` });
-        }
-        /* PDF: pdf-parse varsa metin çıkar (v2 class API — bkz. src/agent/pdf.js) */
-        if (/\.pdf$/i.test(abs)) {
-          let extract = null;
-          try { extract = require('./pdf').extract; } catch {}
-          if (!extract) return JSON.stringify({ ok: false, error: 'pdf okuma yok — npm i pdf-parse' });
-          try {
-            const r = await extract(fs.readFileSync(abs));
-            let content = String((r && r.text) || '').trim();
-            const pages = (r && r.total) || '?';
-            const truncated = content.length > MAX_FILE_CHARS;
-            if (truncated) content = content.slice(0, MAX_FILE_CHARS);
-            if (!content.trim()) return JSON.stringify({ ok: false, error: 'pdf metni çıkarılamadı (taranmış görsel olabilir)' });
-            return JSON.stringify({ ok: true, path: abs, pages, truncated, content });
-          } catch (e2) {
-            return JSON.stringify({ ok: false, error: 'pdf okuma hata: ' + String((e2 && e2.message) || e2) });
-          }
-        }
-        /* binary tespiti: ilk 4KB'ta NUL / %30+ yazdırılamaz karakter (opencode kuralı) */
-        try {
-          const fd = fs.openSync(abs, 'r');
-          let probe;
-          try {
-            probe = Buffer.alloc(Math.min(4096, st.size));
-            fs.readSync(fd, probe, 0, probe.length, 0);
-          } finally {
-            fs.closeSync(fd);
-          }
-          if (looksBinary(probe)) {
-            return JSON.stringify({ ok: false, error: `Cannot read binary file: ${abs}` });
-          }
-        } catch (eProbe) {
-          if (eProbe && /Cannot read binary/.test(String(eProbe.message || ''))) throw eProbe;
-        }
-        let raw;
-        try {
-          raw = readCacheGet(abs, st);
-        } catch {
-          raw = fs.readFileSync(abs, 'utf8');
-        }
-        /* opencode read.ts port: satır numaralı çıktı (`N: içerik`), offset/limit
-           penceresi (1-indexed), 2000 satır varsayılan, uzun satır kırpma */
-        const allLines = raw.split('\n');
-        const offset = Math.max(1, Math.floor(Number(args.offset) || 1));
-        const limit = Math.min(2000, Math.max(1, Math.floor(Number(args.limit) || 2000)));
-        let slice = allLines
-          .slice(offset - 1, offset - 1 + limit)
-          .map((l, i) => {
-            const line = l.length > 2000 ? l.slice(0, 2000) + '... (line truncated to 2000 chars)' : l;
-            return `${offset + i}: ${line}`;
-          });
-        let truncated = offset - 1 + limit < allLines.length;
-        let content = slice.join('\n');
-        /* opencode 50KB byte tavanı: aşarsa satır bazında kes + özel footer */
-        const MAX_BYTES = 50 * 1024;
-        let byteCapped = false;
-        if (Buffer.byteLength(content, 'utf8') > MAX_BYTES) {
-          const keep = [];
-          let used = 0;
-          for (const l of slice) {
-            const need = Buffer.byteLength(l, 'utf8') + 1;
-            if (used + need > MAX_BYTES) break;
-            keep.push(l);
-            used += need;
-          }
-          slice = keep;
-          content = keep.join('\n');
-          truncated = true;
-          byteCapped = keep.length < allLines.length;
-        }
-        /* opencode read.ts footer'ları: byte-kesme → offset devam; satır-kesme →
-           offset devam; aksi → dosya sonu. Model BAŞTAN okuma döngüsüne girmez. */
-        let note;
-        if (byteCapped && slice.length) {
-          note = `(Output capped at 50 KB. Showing lines ${offset}-${offset + slice.length - 1}. Use offset=${offset + slice.length} to continue.)`;
-        } else if (truncated && slice.length) {
-          note = `(Showing lines ${offset}-${offset + slice.length - 1} of ${allLines.length}. Use offset=${offset + slice.length} to continue.)`;
-        } else {
-          note = `(End of file - total ${allLines.length} lines)`;
-        }
-        return JSON.stringify({
-          ok: true,
-          path: abs,
-          totalLines: allLines.length,
-          offset,
-          truncated,
-          note,
-          content,
-        });
-      }
-      case 'write_file': {
-        /* opencode write.ts port: üzerine yazmadan önce eski içerik alınır,
-           sonuçta additions/deletions + UI diffView döner */
-        const abs = safeResolve(String(args.path || ''), cwd);
-        const content = String(args.content ?? '');
-        const existed = fs.existsSync(abs);
-        const before = existed && !fs.statSync(abs).isDirectory() ? fs.readFileSync(abs, 'utf8') : '';
-        fs.mkdirSync(path.dirname(abs), { recursive: true });
-        fs.writeFileSync(abs, content, 'utf8');
-        readCacheDrop(abs); /* sonraki read_file taze içerik okusun */
-        const counts = diffLineCounts(before, content);
-        const out = { ok: true, path: abs, bytes: Buffer.byteLength(content), ...counts };
-        if (ctx.wantDiff) {
-          out.diffView = existed
-            ? { path: abs, ...diffRegion(before, content), ...counts }
-            : { path: abs, before: '', after: capDiffText(content), startLine: 1, ...counts };
-        }
-        return JSON.stringify(out);
-      }
-      case 'list_dir': {
-        /* opencode read.ts dizin modu kuralı: localeCompare sıralı, klasörler
-           `/` ile, 500 giriş tavanı + pagination notu */
-        const abs = args.path ? safeResolve(String(args.path), cwd) : cwd;
-        if (!fs.existsSync(abs)) return JSON.stringify({ ok: false, error: `File not found: ${abs}` });
-        if (!fs.statSync(abs).isDirectory()) return JSON.stringify({ ok: false, error: `Path is a file, not a directory: ${abs}` });
-        const all = fs.readdirSync(abs, { withFileTypes: true });
-        const total = all.length;
-        const entries = all.slice(0, 500);
-        const rows = entries.map((e) => {
-          try {
-            const st = fs.statSync(path.join(abs, e.name));
-            return `${e.isDirectory() ? 'd' : '-'} ${String(st.size).padStart(10)} ${e.isDirectory() ? e.name + '/' : e.name}`;
-          } catch {
-            return `-          ? ${e.name}`;
-          }
-        });
-        rows.sort((a, b) => a.slice(13).localeCompare(b.slice(13)));
-        return JSON.stringify({
-          ok: true,
-          path: abs,
-          count: rows.length,
-          ...(total > 500 ? { note: `(Showing 500 of ${total} entries. Use read_file with offset to paginate.)` } : {}),
-          entries: rows.join('\n'),
-        });
-      }
-      case 'grep': {
-        /* opencode grep.ts BİREBİR port: case-sensitive default (rg kuralı),
-           100 match limit, gruplu çıktı `<abs>:` + `  Line N: text`,
-           2000 karakter satır tavanı + truncation notu */
-        const pattern = String(args.pattern || '');
-        let re;
-        try {
-          re = new RegExp(pattern, args.case_insensitive || args.caseInsensitive ? 'i' : '');
-        } catch (e) {
-          return JSON.stringify({ ok: false, error: 'geçersiz regex: ' + String((e && e.message) || e) });
-        }
-        const root = args.path ? safeResolve(String(args.path), cwd) : cwd;
-        if (!fs.existsSync(root)) return JSON.stringify({ ok: false, error: 'yol bulunamadı: ' + root });
-        const incRe = globToRegExp(String(args.include || ''));
-        const hits = []; // { file, line, text }
-        const MAX_MATCHES = 100;
-        walkFiles(root, incRe, (file) => {
-          if (hits.length >= MAX_MATCHES) return true; // dur
-          let text = '';
-          try {
-            const st = fs.statSync(file);
-            if (st.size > 2 * 1024 * 1024) return false; // dev dosyayı atla
-            text = fs.readFileSync(file, 'utf8');
-          } catch {
-            return false;
-          }
-          const lines = text.split('\n');
-          for (let i = 0; i < lines.length && hits.length < MAX_MATCHES; i++) {
-            if (re.test(lines[i])) {
-              const t = lines[i].slice(0, 2000) + (lines[i].length > 2000 ? '...' : '');
-              hits.push({ file, line: i + 1, text: t });
-            }
-          }
-          return false;
-        });
-        /* opencode çıktı formatı: başlık + dosya gruplama + `  Line N: text` */
-        const groups = new Map();
-        for (const h of hits) {
-          if (!groups.has(h.file)) groups.set(h.file, []);
-          groups.get(h.file).push(`  Line ${h.line}: ${h.text}`);
-        }
-        const out = hits.length
-          ? `Found ${hits.length} matches` +
-            (hits.length >= MAX_MATCHES ? ' (more matches available)' : '') +
-            '\n\n' +
-            [...groups.entries()].map(([f, rows]) => `${f}:\n${rows.join('\n')}`).join('\n\n')
-          : 'No files found';
-        return JSON.stringify({
-          ok: true,
-          pattern,
-          count: hits.length,
-          capped: hits.length >= MAX_MATCHES,
-          ...(hits.length >= MAX_MATCHES ? { note: '(Results truncated. Consider using a more specific path or pattern.)' } : {}),
-          matches: out,
-        });
-      }
-      case 'glob': {
-        /* opencode glob.ts BİREBİR port: 100 sonuç limiti + truncation notu,
-           "No files found", path-dosya hatası */
-        const pat = String(args.pattern || '');
-        if (!pat.trim()) return JSON.stringify({ ok: false, error: 'pattern boş olamaz' });
-        const root = args.path ? safeResolve(String(args.path), cwd) : cwd;
-        if (!fs.existsSync(root)) return JSON.stringify({ ok: false, error: 'yol bulunamadı: ' + root });
-        if (fs.statSync(root).isFile()) {
-          return JSON.stringify({ ok: false, error: `glob path must be a directory: ${root}` });
-        }
-        const re = globToRegExp(pat);
-        const LIMIT = 100;
-        const out = [];
-        walkFiles(root, null, (file) => {
-          const rel = path.relative(root, file).split(path.sep).join('/');
-          if (re.test(rel) || re.test(path.basename(file))) {
-            out.push(path.join(root, rel));
-          }
-          return out.length >= LIMIT; // dur
-        });
-        if (!out.length) return JSON.stringify({ ok: true, pattern: pat, count: 0, files: [], note: 'No files found' });
-        return JSON.stringify({
-          ok: true,
-          pattern: pat,
-          count: out.length,
-          files: out,
-          ...(out.length >= LIMIT ? { note: '(Results are truncated: showing first 100 results. Consider using a more specific path or pattern.)' } : {}),
-        });
-      }
-      case 'web_search': {
-        const q = String(args.query || '');
-        const n = Number(args.max_results) || 8;
-        /* sıralı zincir (obscura/tinyfish/python — dahili tarayıcı engine
-           tarafındaki hook'tan gelir); Exa KALDIRILDI */
-        const r = await searchChainWeb(q, n, { signal: ctx.signal });
-        return JSON.stringify(r);
-      }
-      case 'deep_search': {
-        /* hook'suz bağlamda bile çalışsın: yalnız arama zinciri (sayfa okuma yok).
-           Gizli tarayıcı okuması engine.research hook'undan gelir (main process). */
-        const r = await research.deepSearch(
-          args,
-          { search: (q) => searchChainWeb(q, 10, { signal: ctx.signal }) },
-          ctx.signal
-        );
-        return JSON.stringify(r);
-      }
-      case 'python_run': {
-        const t0 = Date.now();
-        /* önce parametreler — interpreter araması gerektirmeyen hatalar erken dönsün */
-        const cliArgs = Array.isArray(args.args)
-          ? args.args.map(String)
-          : args.args
-            ? [String(args.args)]
-            : [];
-        let scriptPath;
-        let madeTemp = false;
-        if (String(args.code || '').trim()) {
-          scriptPath = path.join(
-            os.tmpdir(),
-            `beast-${Date.now().toString(36)}-${Math.random().toString(36).slice(2, 7)}.py`
-          );
-          fs.writeFileSync(scriptPath, String(args.code), 'utf8');
-          madeTemp = true;
-        } else if (String(args.script || '').trim()) {
-          const raw = String(args.script);
-          scriptPath = path.isAbsolute(raw) ? raw : path.join(pythonScriptsDir(), raw);
-        } else {
-          return JSON.stringify({ ok: false, error: 'code ya da script parametresi gerekli' });
-        }
-        if (!fs.existsSync(scriptPath)) {
-          return JSON.stringify({
-            ok: false,
-            error: `script bulunamadı: ${scriptPath}`,
-            hint: `scripts klasörü: ${pythonScriptsDir()}`,
-          });
-        }
-        let probe;
-        try {
-          probe = await ensurePython({ allowDownload: ctx.allowDownload !== false, signal: ctx.signal });
-        } catch (e) {
-          return JSON.stringify({
-            ok: false,
-            error: String((e && e.message) || e),
-            hint: 'BEAST_PYTHON env ile interpreter yolu verilebilir; ya da python.org kur',
-          });
-        }
-        const r = await runPy(probe.exe, scriptPath, cliArgs, cwd, Number(args.timeout_ms) || 120000, ctx.signal);
-        if (madeTemp) { try { fs.unlinkSync(scriptPath); } catch {} }
-        return JSON.stringify({ ok: r.ok, python: probe.source, exitCode: r.code, ms: Date.now() - t0, output: r.output });
-      }
-      case 'http_fetch':
-      case 'webfetch': {
-        /* opencode webfetch.ts port: format (text|markdown|html) + timeout
-           (default 30s, max 120s) — SSRF koruması Beast'te kalır */
-        const format = String(args.format || 'text').toLowerCase();
-        if (!['text', 'markdown', 'html'].includes(format)) {
-          return JSON.stringify({ ok: false, error: 'format: text | markdown | html' });
-        }
-        const timeoutRaw = Number(args.timeout);
-        if (args.timeout != null && (!Number.isFinite(timeoutRaw) || timeoutRaw <= 0)) {
-          return JSON.stringify({ ok: false, error: `Invalid timeout value: ${args.timeout}. Timeout must be a positive number.` });
-        }
-        /* opencode saniye cinsinden bekler; ms gelirse de tolere et */
-        let timeoutMs = 30000;
-        if (Number.isFinite(timeoutRaw) && timeoutRaw > 0) {
-          timeoutMs = Math.min(timeoutRaw <= 120 ? timeoutRaw * 1000 : timeoutRaw, 120000);
-        }
-        const r = await httpFetch(String(args.url || ''), {
-          maxChars: Number(args.max_chars) || MAX_FETCH_CHARS,
-          format,
-          timeoutMs,
-          signal: ctx.signal,
-        });
-        return JSON.stringify(r);
-      }
-      default:
-        return JSON.stringify({ ok: false, error: `unknown tool ${name}` });
-    }
-  } catch (e) {
-    return JSON.stringify({ ok: false, error: String(e && e.message ? e.message : e) });
-  }
-}
-
-/* ---------- Sıralı arama zinciri (Ayarlar → Web Arama'dan değiştirilir) ----------
-   Motorlar: browser (dahili Chromium → Google) · obscura (stealth headless →
-   DuckDuckGo) · tinyfish (API, anahtar gerekir) · python (ddgs/DDG+Bing+Mojeek).
-   Sıra + aç/kapa ayarı settings.json'da (searchChain) saklanır; Obscura
-   varsayılan AKTİF. Tarayıcı CAPTCHA/trafik verirse 10 dk atlanır. */
-const SEARCH_ENGINE_IDS = ['browser', 'obscura', 'tinyfish', 'python'];
-const DEFAULT_SEARCH_CHAIN = SEARCH_ENGINE_IDS.map((id) => ({ id, on: true }));
-
-let _searchChain = DEFAULT_SEARCH_CHAIN.map((x) => ({ ...x }));
-let _obscuraEnabled = true; /* Obscura varsayılan AKTİF */
-let _browserBanUntil = 0;
-
-function normalizeSearchChain(list) {
-  const arr = Array.isArray(list) ? list : [];
-  const rows = [];
-  const seen = new Set();
-  for (const item of arr) {
-    const id = String((item && item.id) || item || '').trim();
-    if (!SEARCH_ENGINE_IDS.includes(id) || seen.has(id)) continue;
-    seen.add(id);
-    rows.push({ id, on: !(item && item.on === false) });
-  }
-  for (const id of SEARCH_ENGINE_IDS) {
-    if (!seen.has(id)) rows.push({ id, on: true }); /* listede eksik motor varsayılan AÇIK */
-  }
-  /* obscura varsayılan AKTİF; zincir boş kalmasın: hepsi kapalıysa browser'ı aç */
-  if (!rows.some((r) => r.on)) {
-    const b = rows.find((r) => r.id === 'browser');
-    if (b) b.on = true;
-  }
-  return rows;
-}
-
-function setSearchChain(list) {
-  _searchChain = normalizeSearchChain(list);
-  return _searchChain;
-}
-
-function setSearchObscuraEnabled(v) {
-  _obscuraEnabled = v !== false;
-}
-
-function getSearchChain() {
-  return _searchChain.map((x) => ({ ...x }));
-}
-
-function browserBanned() {
-  return Date.now() < _browserBanUntil;
-}
-
-function banBrowser(minutes = 10) {
-  _browserBanUntil = Date.now() + minutes * 60 * 1000;
-}
-
-/* Zinciri sırayla koştur: her motor boş/hata dönerse sıradakine geç.
-   browser motoru yalnız engine'den gelir (hook); araç-bağımsız çağrılarda atlanır. */
-async function searchChainWeb(query, maxResults, { signal, browser } = {}) {
-  const q = String(query || '').trim().slice(0, 400);
-  if (!q) return { ok: false, error: 'boş sorgu' };
-  const cap = Math.max(1, Math.min(12, Number(maxResults) || 8));
-  let out = null;
-  const banned = browserBanned();
-  for (const row of _searchChain) {
-    if (out && out.ok && (out.results || []).length) break;
-    if (!row.on) continue;
-    try {
-      if (row.id === 'browser') {
-        if (typeof browser !== 'function' || banned) continue;
-        out = await browser();
-        if (!out || !out.ok || !(out.results || []).length) {
-          banBrowser(); /* tarayıcı sorunlu — 10 dk atla, alternatifler devrede */
-          out = null;
-        }
-      } else if (row.id === 'obscura') {
-        if (!_obscuraEnabled) continue;
-        out = await obscura.obscuraSearch(q, { maxResults: cap, signal });
-      } else if (row.id === 'tinyfish') {
-        out = await tinyfishSearch(q, cap, signal);
-      } else if (row.id === 'python') {
-        out = await webSearchFast(q, { maxResults: cap, signal });
-      }
-    } catch {
-      out = null;
-    }
-    if (out && out.ok && !(out.results || []).length) out = null; /* boş → sıradaki motor */
-  }
-  if (out && out.ok && banned && (out.results || []).length) {
-    out.note = 'dahili tarayıcı 10 dk askıda (CAPTCHA/trafik) — alternatif motor kullanıldı';
-  }
-  return out || { ok: false, error: 'web arama başarısız — tüm motorlar boş döndü' };
-}
-
-/* ---------- TinyFish (anahtar girilirse zincirdeki kendi sırasında) ----------
-   GET https://api.search.tinyfish.ai?query=...  ·  Header: X-API-Key
-   Anahtar yoksa bu motor otomatik atlanır; sıradaki motor devreye girer. */
-let _tinyfishKey = null;
-
-function setTinyfishKey(key) {
-  _tinyfishKey = String(key || '').trim() || null;
-}
-
-async function tinyfishSearch(query, maxResults, signal) {
-  if (!_tinyfishKey) return null;
-  try {
-    const sig =
-      signal && typeof AbortSignal !== 'undefined' && AbortSignal.any
-        ? AbortSignal.any([signal, AbortSignal.timeout(20000)])
-        : AbortSignal.timeout(20000);
-    const r = await fetch(
-      `https://api.search.tinyfish.ai?query=${encodeURIComponent(String(query || ''))}`,
-      {
-        headers: {
-          'X-API-Key': _tinyfishKey,
-          'X-TF-Request-Origin': 'api',
-          'X-TF-Client-Name': 'tinyfish-api-key-page',
-        },
-        signal: sig,
-      }
-    );
-    if (!r.ok) return null;
-    const j = await r.json();
-    const results = (j.results || [])
-      .map((x) => ({
-        title: String(x.title || ''),
-        url: String(x.url || ''),
-        snippet: String(x.snippet || '').replace(/\s+/g, ' ').slice(0, 400),
-        engine: 'tinyfish',
-      }))
-      .filter((x) => x.url && x.title);
-    if (!results.length) return null;
-    return { ok: true, engine: 'tinyfish', query, results: results.slice(0, Math.max(1, Math.min(12, Number(maxResults) || 8))) };
-  } catch {
-    return null;
-  }
-}
-
-module.exports = {
-  definitions,
-  exec,
-  /* opencode edit.ts portu (test + dış kullanım) */
-  ocReplace,
-  diffLineCounts,
-  diffRegion,
-  ocReplacers: {
-    SimpleReplacer,
-    LineTrimmedReplacer,
-    BlockAnchorReplacer,
-    WhitespaceNormalizedReplacer,
-    IndentationFlexibleReplacer,
-    EscapeNormalizedReplacer,
-    TrimmedBoundaryReplacer,
-    ContextAwareReplacer,
-    MultiOccurrenceReplacer,
-  },
-  runCommand,
-  runShellCommand,
-  runBashCommand,
-  gitbash,
-  disposeShellSessions,
-  boundToolOutput,
-  globToRegExp,
-  walkFiles,
-  assertPublicHttpUrl,
-  parseDdgResults,
-  htmlToText,
-  webSearch,
-  httpFetch,
-  webSearchFast,
-  seedScript,
-  bundledScriptPath,
-  /* arama zinciri (sıra + aç/kapa Ayarlar → Web Arama'dan) */
-  SEARCH_ENGINE_IDS,
-  DEFAULT_SEARCH_CHAIN,
-  setSearchChain,
-  setSearchObscuraEnabled,
-  getSearchChain,
-  searchChainWeb,
-  browserBanned,
-  banBrowser,
-  setTinyfishKey,
-  tinyfishSearch,
-  /* obscura geçişi */
-  obscura,
-  /* python altyapısı */
-  ensurePython,
-  findSystemPython,
-  installEmbeddedPython,
-  pythonScriptsDir,
-  embeddedPythonExe,
-  runPy,
-};
+'use strict';
+
+const fs = require('fs');
+const os = require('os');
+const path = require('path');
+const https = require('https');
+const { execFile } = require('child_process');
+const { spawn } = require('child_process');
+const research = require('./research');
+const searxng = require('./searxng');
+
+const MAX_CMD_OUTPUT = 16000;
+const MAX_FILE_CHARS = 200000;
+
+function truncateMiddle(s, cap) {
+  if (s.length <= cap) return s;
+  const half = Math.floor((cap - 32) / 2);
+  return (
+    s.slice(0, half) +
+    `\n... [${s.length - cap} chars truncated] ...\n` +
+    s.slice(s.length - half)
+  );
+}
+
+/* ---------- opencode port: truncate.ts ----------
+   MAX_LINES 2000 / MAX_BYTES 50KB aşarsa TAM çıktı geçici dosyaya yazılır;
+   modele sınırlı önizleme + dosya yolu döner (Model Tool Output bounding).
+   Böylece hiçbir araç çıktısı KAYBOLMAZ — ajan read_file ile kalanını okur. */
+const TRUNC_MAX_LINES = 2000;
+const TRUNC_MAX_BYTES = 50 * 1024;
+const TRUNC_PREVIEW_CHARS = 6000; /* engine'in 7200'lik dilimi içinde ipucu kalsın */
+
+function toolOutputDir() {
+  const d = path.join(os.tmpdir(), 'beast-tool-output');
+  try { fs.mkdirSync(d, { recursive: true }); } catch {}
+  return d;
+}
+
+function boundToolOutput(text) {
+  const s = String(text || '');
+  const bytes = Buffer.byteLength(s, 'utf8');
+  const lines = s.split('\n').length;
+  if (bytes <= TRUNC_MAX_BYTES && lines <= TRUNC_MAX_LINES) return { text: s };
+  const file = path.join(
+    toolOutputDir(),
+    `tool_${Date.now().toString(36)}_${Math.random().toString(36).slice(2, 7)}.txt`
+  );
+  try {
+    fs.writeFileSync(file, s);
+  } catch {
+    return { text: s.slice(0, TRUNC_PREVIEW_CHARS) + '\n…[kırpıldı]' };
+  }
+  return {
+    text:
+      s.slice(0, TRUNC_PREVIEW_CHARS) +
+      `\n\n[çıktı kırpıldı: ${lines} satır / ${bytes} byte — TAM ÇIKTI: ${file}\nread_file'ı offset/limit ile kullanarak kalan bölümleri oku]`,
+    outputFile: file,
+  };
+}
+
+/* ---------- opencode port: shell.ts + shell/prompt.ts ----------
+   Workspace başına KALICI PowerShell oturumu: cd/env çağrılar arasında
+   korunur, her çağrıda process spawn maliyeti yoktur. Komut bitişi benzersiz
+   işaretle algılanır (prompt senkronizasyonu). Paralel çağrılar oturum
+   başına sıraya girer (opencode ile aynı: oturum başına sıralı yürütme). */
+const SHELL_QUEUE_CAP = 6; /* en fazla bu kadar ayrı workspace oturumu */
+const SHELL_IDLE_MS = 10 * 60 * 1000; /* 10 dk boşta kalan oturum kapatılır */
+const _shSessions = new Map(); // cwd → session
+
+/* boşta reaper: unref'li zamanlayıcı — event loop'u tek başına tutmaz */
+const _shellReaper = setInterval(() => {
+  const now = Date.now();
+  for (const [k, sess] of _shSessions) {
+    if (!sess.busy && now - (sess.lastUsed || 0) > SHELL_IDLE_MS) {
+      _shellDispose(sess);
+      _shSessions.delete(k);
+    }
+  }
+}, 60000);
+if (_shellReaper.unref) _shellReaper.unref();
+
+function _shellDispose(sess) {
+  try {
+    if (sess.proc && sess.proc.pid) {
+      spawn('taskkill', ['/pid', String(sess.proc.pid), '/T', '/F'], { windowsHide: true });
+    }
+  } catch {}
+  try { sess.proc && sess.proc.kill(); } catch {}
+  sess.dead = true;
+}
+
+function disposeShellSessions() {
+  for (const sess of _shSessions.values()) _shellDispose(sess);
+  _shSessions.clear();
+}
+
+function _spawnShell(cwd) {
+  const proc = spawn(
+    'powershell.exe',
+    ['-NoProfile', '-NoExit', '-Command', '-'],
+    { cwd: cwd || process.cwd(), windowsHide: true, env: envWithPathPrefix() }
+  );
+  const sess = {
+    proc,
+    cwd,
+    buf: '',
+    err: '',
+    seq: 0,
+    busy: false,
+    queue: [], // {command, resolve, timer, signal, onAbort}
+    dead: false,
+  };
+  /* Node 22: child + stdio stream'leri event loop'a bağlanmaz — test/CLI'da
+     bekleyen oturum sürecin çıkmasını ENGELLEMEZ; uygulama kapanışında
+     disposeShellSessions() yine de temiz kapatır */
+  try {
+    proc.unref && proc.unref();
+    proc.stdin.unref && proc.stdin.unref();
+    proc.stdout.unref && proc.stdout.unref();
+    proc.stderr.unref && proc.stderr.unref();
+  } catch {}
+  proc.stdin.on && proc.stdin.on('error', () => {}); // EPIPE yut — oturum ölünce yazma patlamasın
+  proc.stdout.setEncoding('utf8');
+  proc.stderr.setEncoding('utf8');
+  proc.stdout.on('data', (d) => {
+    sess.buf += d;
+    _shellPump(sess);
+  });
+  proc.stderr.on('data', (d) => {
+    sess.err += d;
+    if (sess.err.length > MAX_CMD_OUTPUT * 4) sess.err = sess.err.slice(-MAX_CMD_OUTPUT * 2);
+  });
+  proc.on('exit', () => {
+    sess.dead = true;
+    /* bekleyen komutları ölü oturumda düşür — yeni çağrı taze oturum açar */
+    while (sess.queue.length) {
+      const w = sess.queue.shift();
+      clearTimeout(w.timer);
+      w.resolve({ ok: false, code: null, output: sess.buf.slice(-2000) + '\n[beast] shell oturumu kapandı' });
+    }
+    sess.busy = false;
+    if (_shSessions.get(cwd) === sess) _shSessions.delete(cwd);
+  });
+  proc.on('error', () => {
+    sess.dead = true;
+  });
+  return sess;
+}
+
+function _shellPump(sess) {
+  const w = sess.queue[0];
+  if (!w || !w.dispatched) return;
+  const marker = `${SHELL_MARKER_PREFIX}${w.n}_`;
+  const mi = sess.buf.indexOf(marker);
+  if (mi < 0) return;
+  const lineEnd = sess.buf.indexOf('\n', mi);
+  if (lineEnd < 0) return; /* işaret satırı tam gelmedi — daha fazla veri bekle */
+  const markerLine = sess.buf.slice(mi, lineEnd).trim();
+  let out = sess.buf.slice(0, mi);
+  sess.buf = sess.buf.slice(lineEnd + 1);
+  sess.queue.shift();
+  sess.busy = sess.queue.length > 0;
+  clearTimeout(w.timer);
+  const codeM = /_(\-?\d+)\s*$/.exec(markerLine);
+  const exitCode = codeM ? Number(codeM[1]) : null;
+  if (out.length > MAX_CMD_OUTPUT * 4) out = out.slice(-MAX_CMD_OUTPUT * 2);
+  const errPart = sess.err.trim();
+  sess.err = '';
+  w.resolve({
+    ok: exitCode === 0,
+    code: exitCode,
+    output:
+      (out.trim() || '') +
+      (errPart ? (out.trim() ? '\n[stderr] ' : '') + errPart : ''),
+  });
+  sess.lastUsed = Date.now();
+  if (sess.queue.length) _shellDispatch(sess);
+}
+
+const SHELL_MARKER_PREFIX = '__BEAST_DONE_';
+
+function _shellDispatch(sess) {
+  const w = sess.queue[0];
+  if (!w || w.dispatched) return;
+  w.dispatched = true;
+  w.n = ++sess.seq;
+  const cmd = w.command;
+  const marker = `${SHELL_MARKER_PREFIX}${w.n}_$LASTEXITCODE`;
+  /* $LASTEXITCODE sıfırlanır: yalnız cmdlet koşan komutlar da ok:true dönsün;
+     native exe hata verirse gerçek kod işaret satırına yazılır */
+  sess.proc.stdin.write(`$global:LASTEXITCODE = 0\n${cmd}${cmd.endsWith('\n') ? '' : '\n'}Write-Output "${marker}"\n`);
+  const finishTimeout = () => {
+    _shellDispose(sess);
+    if (_shSessions.get(sess.cwd) === sess) _shSessions.delete(sess.cwd);
+    sess.busy = false;
+    while (sess.queue.length) {
+      const x = sess.queue.shift();
+      clearTimeout(x.timer);
+      x.resolve({ ok: false, code: null, output: '[beast] shell komutu zaman aşımı — oturum tazelendi' });
+    }
+  };
+  w.timer = setTimeout(finishTimeout, w.timeoutMs);
+  if (w.signal) {
+    if (w.signal.aborted) return finishTimeout();
+    w.signal.addEventListener('abort', finishTimeout, { once: true });
+  }
+}
+
+function runShellCommand(command, cwd, timeoutMs = 90000, signal) {
+  return new Promise((resolve) => {
+    let sess = _shSessions.get(cwd);
+    if (!sess || sess.dead) {
+      try {
+        sess = _spawnShell(cwd);
+        _shSessions.set(cwd, sess);
+        while (_shSessions.size > SHELL_QUEUE_CAP) {
+          const [k, old] = _shSessions.entries().next().value;
+          if (k === cwd) break;
+          _shellDispose(old);
+          _shSessions.delete(k);
+        }
+      } catch {
+        /* kalıcı oturum açılamadı → tek seferlik klasik yol */
+        return runCommand(command, cwd, timeoutMs, signal).then(resolve);
+      }
+    }
+    sess.queue.push({ command, resolve, timeoutMs, signal, dispatched: false });
+    sess.lastUsed = Date.now();
+    if (!sess.busy) {
+      sess.busy = true;
+      _shellDispatch(sess);
+    }
+  });
+}
+
+/* --- core/shell.ts gitbash() port: Windows'ta bash çözümleme sırası --- */
+function gitbash() {
+  const cands = [];
+  if (process.env.ProgramFiles) cands.push(path.join(process.env.ProgramFiles, 'Git', 'bin', 'bash.exe'));
+  if (process.env['ProgramFiles(x86)']) cands.push(path.join(process.env['ProgramFiles(x86)'], 'Git', 'bin', 'bash.exe'));
+  if (process.env.LOCALAPPDATA) cands.push(path.join(process.env.LOCALAPPDATA, 'Programs', 'Git', 'bin', 'bash.exe'));
+  for (const c of cands) {
+    try { if (fs.existsSync(c)) return c; } catch {}
+  }
+  return 'bash'; /* PATH'te aranır; yoksa spawn ENOENT → net hata mesajı */
+}
+
+function runBashCommand(command, cwd, timeoutMs = 90000, signal) {
+  return new Promise((resolve) => {
+    let settled = false;
+    let out = '';
+    let err = '';
+    const finish = (code, killed) => {
+      if (settled) return;
+      settled = true;
+      clearTimeout(timer);
+      const text = (out ? out.trim() : '') + (err ? (out ? '\n[stderr] ' : '') + err.trim() : '');
+      resolve({ ok: !killed && code === 0, code, output: truncateMiddle(text || '(no output)', MAX_CMD_OUTPUT) });
+    };
+    let child;
+    try {
+      child = spawn(gitbash(), ['-lc', command], {
+        cwd: cwd || process.cwd(),
+        windowsHide: true,
+        env: envWithPathPrefix(),
+      });
+    } catch {
+      resolve({ ok: false, code: null, output: 'bash bulunamadı — Git for Windows kur ya da PowerShell kullan' });
+      return;
+    }
+    const timer = setTimeout(() => {
+      try { spawn('taskkill', ['/pid', String(child.pid), '/T', '/F'], { windowsHide: true }); } catch {}
+      err += '\n[beast] command timed out';
+      finish(null, true);
+    }, timeoutMs);
+    if (signal) {
+      if (signal.aborted) return finish(null, true);
+      signal.addEventListener('abort', () => {
+        try { spawn('taskkill', ['/pid', String(child.pid), '/T', '/F'], { windowsHide: true }); } catch {}
+        finish(null, true);
+      }, { once: true });
+    }
+    child.stdout.setEncoding('utf8');
+    child.stderr.setEncoding('utf8');
+    child.stdout.on('data', (d) => { out += d; });
+    child.stderr.on('data', (d) => { err += d; });
+    child.on('error', (e) => {
+      clearTimeout(timer);
+      settled = true;
+      const msg = String((e && e.message) || '');
+      resolve({
+        ok: false,
+        code: null,
+        output: /ENOENT/i.test(msg)
+          ? 'bash bulunamadı — Git for Windows kur ya da PowerShell kullan'
+          : msg,
+      });
+    });
+    child.on('close', (code) => finish(code, false));
+  });
+}
+
+function runCommand(command, cwd, timeoutMs = 90000, signal) {
+  return new Promise((resolve) => {
+    let settled = false;
+    let out = '';
+    let err = '';
+    const finish = (code, killed) => {
+      if (settled) return;
+      settled = true;
+      clearTimeout(timer);
+      const text =
+        (out ? out.trim() : '') +
+        (err ? (out ? '\n[stderr] ' : '') + err.trim() : '');
+      resolve({
+        ok: !killed && code === 0,
+        output: truncateMiddle(text || '(no output)', MAX_CMD_OUTPUT),
+        code,
+      });
+    };
+
+    const child = spawn(
+      'powershell.exe',
+      ['-NoProfile', '-NonInteractive', '-ExecutionPolicy', 'Bypass', '-Command', command],
+      { cwd: cwd || process.cwd(), windowsHide: true, env: envWithPathPrefix() }
+    );
+
+    const timer = setTimeout(() => {
+      try {
+        spawn('taskkill', ['/pid', String(child.pid), '/T', '/F'], { windowsHide: true });
+      } catch {}
+      finish(null, true);
+      // append notice
+      err += '\n[beast] command timed out';
+    }, timeoutMs);
+
+    if (signal) {
+      if (signal.aborted) return finish(null, true);
+      signal.addEventListener('abort', () => {
+        try {
+          spawn('taskkill', ['/pid', String(child.pid), '/T', '/F'], { windowsHide: true });
+        } catch {}
+        finish(null, true);
+      }, { once: true });
+    }
+
+    child.stdout.setEncoding('utf8');
+    child.stderr.setEncoding('utf8');
+    child.stdout.on('data', (d) => {
+      out += d;
+      if (out.length > MAX_CMD_OUTPUT * 4) out = out.slice(-MAX_CMD_OUTPUT * 2);
+    });
+    child.stderr.on('data', (d) => {
+      err += d;
+      if (err.length > MAX_CMD_OUTPUT * 4) err = err.slice(-MAX_CMD_OUTPUT * 2);
+    });
+    child.on('error', (e) => finish(1, false) || void (err += String(e.message)));
+    child.on('close', (code) => finish(code, false));
+  });
+}
+
+function safeResolve(p, cwd) {
+  const abs = path.isAbsolute(p) ? p : path.join(cwd || process.cwd(), p);
+  return path.normalize(abs);
+}
+
+/* ---------- opencode read.ts port: binary tespit + fuzzy öneri ----------
+   opencode read.ts kuralı: uzantı kara listesi YA DA ilk 4KB'ta NUL byte YA DA
+   %30+'ı yazdırılamaz karakter → dosya binary sayılır, okunmaz. */
+const BINARY_EXT_RE =
+  /\.(zip|tar|gz|tgz|bz2|xz|7z|rar|exe|dll|so|dylib|bin|iso|img|msi|apk|jar|class|pyc|pyo|o|obj|a|lib|woff2?|ttf|otf|eot|mp3|mp4|avi|mkv|mov|flac|ogg|wav|webm|psd|ai|sketch|db|sqlite3?|pdb|docx?|xlsx?|pptx?|odt|ods|odp|pgp|gpg|keystore|jks|p12|traineddata|idx)$/i;
+
+function looksBinary(buf) {
+  const sample = buf.length > 4096 ? buf.subarray(0, 4096) : buf;
+  if (sample.includes(0)) return true;
+  let nonPrintable = 0;
+  for (const b of sample) {
+    if (b === 9 || b === 10 || b === 13) continue; // \t \n \r
+    if (b < 32 || b === 127 || b >= 0x80) nonPrintable++; // 0x80+ UTF-8 devam byte'ı olabilir ama kaba tarama yeterli
+  }
+  return sample.length > 0 && nonPrintable / sample.length > 0.3;
+}
+
+/* opencode read.ts: dosya yoksa aynı klasörde isim ön-ek benzerliği olan 3 kardeş öner */
+function fuzzySiblings(abs) {
+  try {
+    const dir = path.dirname(abs);
+    const base = path.basename(abs).toLowerCase();
+    const prefix = base.slice(0, 4);
+    const score = (name) => {
+      const n = name.toLowerCase();
+      if (n === base) return -1;
+      let s = 0;
+      if (n.startsWith(prefix)) s += 2;
+      for (let i = 0; i < Math.min(base.length, n.length); i++) if (base[i] === n[i]) s += 0.1;
+      return s;
+    };
+    return fs
+      .readdirSync(dir)
+      .map((name) => ({ name, s: score(name) }))
+      .filter((x) => x.s > 0)
+      .sort((a, b) => b.s - a.s)
+      .slice(0, 3)
+      .map((x) => x.name);
+  } catch {
+    return [];
+  }
+}
+
+/* ---------- opencode port: grep/glob altyapısı ----------
+   ripgrep gitignore'u izler; Beast'te bağımlılık olmasın diye ağır klasörler
+   statik atlanır (node_modules, .git, derleme çıktıları…). */
+const SKIP_DIRS = new Set([
+  'node_modules', '.git', '__pycache__', '.venv', 'venv', 'env',
+  'dist', 'build', 'out', 'coverage', '.next', '.turbo', '.cache',
+]);
+
+/* mini glob → regex: ** → her şey, * → / içermeyen her şey, ? → tek karakter */
+function globToRegExp(glob) {
+  const g = String(glob || '').trim();
+  if (!g) return null;
+  let re = '';
+  for (let i = 0; i < g.length; i++) {
+    const c = g[i];
+    if (c === '*') {
+      if (g[i + 1] === '*') {
+        re += '.*';
+        i++;
+        if (g[i + 1] === '/') i++; /* yıldız-yıldız-slash: üst klasörler opsiyonel */
+      } else re += '[^/\\\\]*';
+    } else if (c === '?') re += '[^/\\\\]';
+    else re += c.replace(/[.+^${}()|[\]\\]/g, '\\$&');
+  }
+  try {
+    return new RegExp('^' + re + '$', 'i');
+  } catch {
+    return null;
+  }
+}
+
+/* Sıralı dosya yürüyüşü; cb true dönerse erken durur */
+function walkFiles(dir, includeRe, cb, depth = 0) {
+  if (depth > 24) return false;
+  let entries = [];
+  try {
+    entries = fs.readdirSync(dir, { withFileTypes: true });
+  } catch {
+    return false;
+  }
+  for (const e of entries) {
+    const full = path.join(dir, e.name);
+    if (e.isDirectory()) {
+      if (SKIP_DIRS.has(e.name) || e.name.startsWith('.') && e.name !== '.opencode') continue;
+      if (walkFiles(full, includeRe, cb, depth + 1)) return true;
+    } else if (e.isFile()) {
+      if (includeRe && !includeRe.test(e.name)) continue;
+      if (cb(full)) return true;
+    }
+  }
+  return false;
+}
+
+/* ---------- Python altyapısı (#18) ----------
+   Sıra: BEAST_PYTHON env → sistem python/python3/py -3 → gömülü runtime
+   (%APPDATA%\beast\py — yoksa TEK SEFERLİK kendisi indirip açar).
+   Böylece ajanlar makinede Python kurulmasa bile python_run kullanabilir. */
+
+const PYTHON_EMBED_URL = 'https://www.python.org/ftp/python/3.12.8/python-3.12.8-embed-amd64.zip';
+const PY_PROBE_TIMEOUT = 6000;
+
+function beastAppDir() {
+  const base = process.env.APPDATA || path.join(os.homedir(), 'AppData', 'Roaming');
+  return path.join(base, 'beast');
+}
+
+function pythonScriptsDir() {
+  const d = path.join(beastAppDir(), 'scripts');
+  fs.mkdirSync(d, { recursive: true });
+  return d;
+}
+
+function embeddedPythonExe() {
+  return path.join(beastAppDir(), 'py', 'python.exe');
+}
+
+/* Gömülü runtime kuruluysa klasörünü PATH önüne koy: ajanın run_command'ında
+   `python` / `python.exe` da çalışsın (makinede Python kurulmuş olmasa bile). */
+function pythonPathPrefix() {
+  try {
+    const d = path.dirname(embeddedPythonExe());
+    if (fs.existsSync(path.join(d, 'python.exe'))) return d + path.delimiter;
+  } catch {}
+  return '';
+}
+
+function envWithPathPrefix() {
+  const base = { ...process.env };
+  base.PATH = pythonPathPrefix() + (process.env.PATH || process.env.Path || '');
+  return base;
+}
+
+/* paketlenmiş python scriptleri (src/agent/scripts) → kullanıcı scripts klasörüne tohumla */
+function bundledScriptPath(name) {
+  return path.join(__dirname, 'scripts', name);
+}
+
+function seedScript(name, force = false) {
+  const src = bundledScriptPath(name);
+  try {
+    const dest = path.join(pythonScriptsDir(), name);
+    /* force: paket içindeki güncel sürüm her zaman kazanır (sistem scripti) */
+    if (fs.existsSync(src) && (force || !fs.existsSync(dest))) {
+      fs.copyFileSync(src, dest);
+    }
+    return dest;
+  } catch {
+    return path.join(pythonScriptsDir(), name);
+  }
+}
+
+/* #20 hızlı web arama: ddgs kütüphanesi varsa onu kullan; yoksa python çoklu-motor
+   paralel; o da olmazsa JS fallback. Sonuç boş gelirse (CAPTCHA/engel) ddgs'yi
+   TEK SEFERLİK arka planda kurup tekrar dener. */
+function pyExec(exe, args, timeoutMs) {
+  return new Promise((resolve) => {
+    try {
+      execFile(exe, args, { timeout: timeoutMs, windowsHide: true }, (err, stdout, stderr) =>
+        resolve({ err, out: String(stdout || '') + String(stderr || '') })
+      );
+    } catch (e) {
+      resolve({ err: e, out: '' });
+    }
+  });
+}
+
+let _ddgsTried = null; // Promise<boolean> — oturum başına tek kurulum denemesi
+
+function ddgsImportOk(probe) {
+  /* hızlı yerel kontrol: kütüphane import edilebiliyor mu (ağ gerektirmez) */
+  return pyExec(probe.exe, ['-c', 'import ddgs'], 15000).then((r) => !r.err);
+}
+
+/* pip ile TEK SEFERLİK kurulum dener; uzun sürebilir — await ETME, arka planda bırak */
+function ensureDdgs(probe) {
+  if (_ddgsTried) return _ddgsTried;
+  _ddgsTried = (async () => {
+    try {
+      if (await ddgsImportOk(probe)) return true;
+      const pip = await pyExec(probe.exe, ['-m', 'pip', '--version'], 15000);
+      if (pip.err) return false; // gömülü python'da pip yok — sessizce vazgeç
+      const inst = await pyExec(
+        probe.exe,
+        ['-m', 'pip', 'install', '--quiet', '--disable-pip-version-check', 'ddgs'],
+        120000
+      );
+      if (inst.err) return false;
+      return await ddgsImportOk(probe);
+    } catch {
+      return false;
+    }
+  })();
+  return _ddgsTried;
+}
+
+function parseWsOutput(output) {
+  return String(output || '')
+    .split('\n')
+    .map((l) => l.trim())
+    .filter((l) => l.startsWith('{'))
+    .map((l) => {
+      try { return JSON.parse(l); } catch { return null; }
+    })
+    .filter(Boolean)
+    .map((x) => ({ title: String(x.title || ''), url: String(x.url || ''), snippet: String(x.snippet || ''), engine: String(x.engine || '') }));
+}
+
+/* Obscura yaklaşımının Node-gömülü uyarlaması: Chrome TLS parmak izi taklidi.
+   curl_cffi (curl-impersonate) gömülü Python'a otomatik kurulur; ayrı program YOK.
+   Bot koruması olan sitelerden (DDG html) gerçek Chrome kimliğiyle sonuç çeker. */
+
+let _cffiTried = null; // Promise<boolean> — oturum başına tek kurulum denemesi
+
+function cffiImportOk(probe) {
+  return pyExec(probe.exe, ['-c', 'import curl_cffi'], 15000).then((r) => !r.err);
+}
+
+function ensureCffi(probe) {
+  if (_cffiTried) return _cffiTried;
+  _cffiTried = (async () => {
+    try {
+      if (await cffiImportOk(probe)) return true;
+      const pip = await pyExec(probe.exe, ['-m', 'pip', '--version'], 15000);
+      if (pip.err) return false;
+      const inst = await pyExec(
+        probe.exe,
+        ['-m', 'pip', 'install', '--quiet', '--disable-pip-version-check', 'curl_cffi'],
+        300000
+      );
+      if (inst.err) return false;
+      return await cffiImportOk(probe);
+    } catch {
+      return false;
+    }
+  })();
+  return _cffiTried;
+}
+
+/* stealth arama: curl_cffi ile DDG html — ham HTML döner, JS'te parse edilir */
+async function stealthSearch(query, { maxResults = 8, signal } = {}) {
+  try {
+    const probe = await ensurePython({ allowDownload: true, signal });
+    if (!(await cffiImportOk(probe))) {
+      ensureCffi(probe).catch(() => {}); /* ilk çağrıda atla, kurulum arka planda */
+      return null;
+    }
+    const script = seedScript('stealthsearch.py', true);
+    const r = await runPy(
+      probe.exe,
+      script,
+      [encodeURIComponent(String(query || ''))],
+      process.cwd(),
+      25000,
+      signal
+    );
+    const html = String(r.output || '');
+    if (!html || html.length < 500) return null;
+    const results = parseDdgResults(html, Math.min(Math.max(Number(maxResults) || 8, 1), 12));
+    if (!results.length) return null;
+    return {
+      ok: true,
+      engine: 'stealth',
+      query,
+      results: results.map((x) => ({ ...x, engine: 'stealth' })),
+    };
+  } catch {
+    return null;
+  }
+}
+
+async function webSearchFast(query, { maxResults = 8, signal } = {}) {
+  try {
+    const probe = await ensurePython({ allowDownload: true, signal });
+    const script = seedScript('websearch.py', true);
+    let r = await runPy(probe.exe, script, ['--json', '--limit', String(maxResults), query], process.cwd(), 20000, signal);
+    let results = parseWsOutput(r.output);
+    if (!results.length) {
+      /* muhtemel CAPTCHA/engel — ddgs kuruluysa hemen onunla tekrar dene;
+         kurulu değilse kurulumu arka plana bırak (bu çağrıyı bekletmez) */
+      if (await ddgsImportOk(probe)) {
+        r = await runPy(probe.exe, script, ['--json', '--limit', String(maxResults), query], process.cwd(), 20000, signal);
+        results = parseWsOutput(r.output);
+      } else {
+        ensureDdgs(probe).catch(() => {});
+      }
+    } else if (!results.some((x) => x.engine === 'ddgs')) {
+      /* arama çalışıyor ama ddgs henüz yok — arka planda sessiz kur, sonraki aramalar hızlanır */
+      ensureDdgs(probe).catch(() => {});
+    }
+    if (results.length) {
+      return { ok: true, engine: results.some((x) => x.engine === 'ddgs') ? 'ddgs' : 'python-multi', query, results };
+    }
+  } catch {}
+  /* JS fallback (tek motor DDG) */
+  return webSearch(query, { maxResults, signal });
+}
+
+let _pyProbe = null;
+
+function probeInterpreter(exe, prefix) {
+  return new Promise((resolve) => {
+    try {
+      execFile(exe, [...(prefix || []), '--version'], { timeout: PY_PROBE_TIMEOUT, windowsHide: true }, (err, stdout, stderr) => {
+        if (err) return resolve(null);
+        const v = String(stdout || stderr || '').trim();
+        resolve(/Python\s*3/i.test(v) ? { exe, prefix: prefix || [], source: 'system', version: v } : null);
+      });
+    } catch {
+      resolve(null);
+    }
+  });
+}
+
+async function findSystemPython() {
+  if (process.env.BEAST_PYTHON && fs.existsSync(process.env.BEAST_PYTHON)) {
+    return { exe: process.env.BEAST_PYTHON, prefix: [], source: 'env' };
+  }
+  for (const cand of [
+    ['python.exe', []],
+    ['python3.exe', []],
+    ['py.exe', ['-3']],
+  ]) {
+    const r = await probeInterpreter(cand[0], cand[1]);
+    if (r) return r;
+  }
+  return null;
+}
+
+function httpsGetBuffer(url, redirectsLeft = 4, signal) {
+  return new Promise((resolve, reject) => {
+    const req = https.get(url, { headers: { 'User-Agent': 'BeastAgent/1.0 (+python bootstrap)' } }, (res) => {
+      if ([301, 302, 303, 307, 308].includes(res.statusCode) && res.headers.location && redirectsLeft > 0) {
+        res.resume();
+        return resolve(httpsGetBuffer(new URL(res.headers.location, url).toString(), redirectsLeft - 1, signal));
+      }
+      if (res.statusCode !== 200) {
+        res.resume();
+        return reject(new Error(`indirme başarısız: HTTP ${res.statusCode}`));
+      }
+      const chunks = [];
+      let size = 0;
+      res.on('data', (c) => {
+        size += c.length;
+        if (size > 80 * 1024 * 1024) {
+          req.destroy(new Error('dosya çok büyük'));
+          return;
+        }
+        chunks.push(c);
+      });
+      res.on('end', () => resolve(Buffer.concat(chunks)));
+      res.on('error', reject);
+    });
+    req.on('error', reject);
+    if (signal) {
+      if (signal.aborted) return reject(new Error('iptal'));
+      signal.addEventListener('abort', () => req.destroy(new Error('iptal')), { once: true });
+    }
+  });
+}
+
+async function installEmbeddedPython(signal) {
+  const dest = path.join(beastAppDir(), 'py');
+  fs.mkdirSync(dest, { recursive: true });
+  const zipPath = path.join(os.tmpdir(), 'beast-py-embed.zip');
+  const buf = await httpsGetBuffer(PYTHON_EMBED_URL, 4, signal);
+  fs.writeFileSync(zipPath, buf);
+  const r = await runCommand(
+    `Expand-Archive -LiteralPath "${zipPath}" -DestinationPath "${dest}" -Force`,
+    null,
+    180000
+  );
+  try { fs.unlinkSync(zipPath); } catch {}
+  if (!r.ok || !fs.existsSync(embeddedPythonExe())) {
+    throw new Error('gömülü python kurulamadı: ' + String(r.output || '').slice(0, 200));
+  }
+  /* site-packages yolunu aç — pip ile kurulabilen paketler için ön hazırlık */
+  try {
+    const pth = path.join(dest, 'python312._pth');
+    if (fs.existsSync(pth)) {
+      const cur = fs.readFileSync(pth, 'utf8');
+      if (!/import site/.test(cur)) fs.writeFileSync(pth, cur.replace(/\n*$/, '\nimport site\n'), 'utf8');
+    }
+  } catch {}
+  return { exe: embeddedPythonExe(), prefix: [], source: 'embedded', version: '3.12.8 (gömülü)' };
+}
+
+async function ensurePython({ allowDownload = true, signal } = {}) {
+  if (_pyProbe) {
+    try { fs.accessSync(_pyProbe.exe); return _pyProbe; } catch { _pyProbe = null; }
+  }
+  const emb = embeddedPythonExe();
+  if (fs.existsSync(emb)) {
+    _pyProbe = { exe: emb, prefix: [], source: 'embedded' };
+    return _pyProbe;
+  }
+  const sys = await findSystemPython();
+  if (sys) {
+    _pyProbe = sys;
+    return sys;
+  }
+  if (!allowDownload) throw new Error('python bulunamadı ve otomatik kurulum kapalı');
+  _pyProbe = await installEmbeddedPython(signal);
+  return _pyProbe;
+}
+
+function runPy(exe, scriptPath, cliArgs, cwd, timeoutMs = 120000, signal) {
+  return new Promise((resolve) => {
+    let settled = false;
+    let out = '';
+    let err = '';
+    const startedAt = Date.now();
+    const finish = (code, killed) => {
+      if (settled) return;
+      settled = true;
+      clearTimeout(timer);
+      const text =
+        (out ? out.trim() : '') +
+        (err ? (out ? '\n[stderr] ' : '') + err.trim() : '');
+      resolve({
+        ok: !killed && code === 0,
+        code,
+        ms: Date.now() - startedAt,
+        output: truncateMiddle(text || '(no output)', MAX_CMD_OUTPUT),
+      });
+    };
+
+    let child;
+    try {
+      child = spawn(exe, [scriptPath, ...cliArgs], {
+        cwd: cwd || process.cwd(),
+        windowsHide: true,
+        env: { ...envWithPathPrefix(), PYTHONUTF8: '1', PYTHONIOENCODING: 'utf-8' },
+      });
+    } catch (e) {
+      err += String(e.message);
+      return finish(1, false);
+    }
+
+    const timer = setTimeout(() => {
+      try { spawn('taskkill', ['/pid', String(child.pid), '/T', '/F'], { windowsHide: true }); } catch {}
+      finish(null, true);
+      err += '\n[beast] python timed out';
+    }, timeoutMs);
+
+    if (signal) {
+      if (signal.aborted) return finish(null, true);
+      signal.addEventListener('abort', () => {
+        try { spawn('taskkill', ['/pid', String(child.pid), '/T', '/F'], { windowsHide: true }); } catch {}
+        finish(null, true);
+      }, { once: true });
+    }
+
+    child.stdout.setEncoding('utf8');
+    child.stderr.setEncoding('utf8');
+    child.stdout.on('data', (d) => {
+      out += d;
+      if (out.length > MAX_CMD_OUTPUT * 4) out = out.slice(-MAX_CMD_OUTPUT * 2);
+    });
+    child.stderr.on('data', (d) => {
+      err += d;
+      if (err.length > MAX_CMD_OUTPUT * 4) err = err.slice(-MAX_CMD_OUTPUT * 2);
+    });
+    child.on('error', (e) => { err += String(e.message); finish(1, false); });
+    child.on('close', (code) => finish(code, false));
+  });
+}
+
+/* ---------- web araçları ---------- */
+
+const UA =
+  'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126 Safari/537.36 BeastAgent/0.2';
+
+const PRIVATE_HOST_RE =
+  /^(localhost|.*\.local|127\.|10\.|192\.168\.|169\.254\.|0\.|::1|\[::1\]|f[cd][0-9a-f]{2}:)/i;
+
+function assertPublicHttpUrl(raw) {
+  let u;
+  try {
+    u = new URL(String(raw || ''));
+  } catch {
+    throw new Error('geçersiz URL');
+  }
+  if (u.protocol !== 'http:' && u.protocol !== 'https:') {
+    throw new Error('yalnızca http/https desteklenir');
+  }
+  const host = u.hostname;
+  if (PRIVATE_HOST_RE.test(host)) {
+    throw new Error('yerel/ağ içi adreslere erişim engellendi');
+  }
+  if (/^172\.(1[6-9]|2\d|3[01])\./.test(host)) {
+    throw new Error('yerel/ağ içi adreslere erişim engellendi');
+  }
+  if ((u.username || u.password)) {
+    throw new Error('URL içinde kullanıcı bilgisi desteklenmez');
+  }
+  return u.toString();
+}
+
+const ENTITIES = {
+  amp: '&', lt: '<', gt: '>', quot: '"', apos: "'", nbsp: ' ', mdash: '—', ndash: '–',
+  hellip: '…', rsquo: '\u2019', lsquo: '\u2018', ldquo: '\u201c', rdquo: '\u201d',
+};
+
+function decodeEntities(s) {
+  return String(s || '')
+    .replace(/&#x([0-9a-f]+);/gi, (_m, h) => {
+      try { return String.fromCodePoint(parseInt(h, 16)); } catch { return ''; }
+    })
+    .replace(/&#(\d+);/g, (_m, d) => {
+      try { return String.fromCodePoint(Number(d)); } catch { return ''; }
+    })
+    .replace(/&([a-z]+);/gi, (m, name) => ENTITIES[name.toLowerCase()] ?? m);
+}
+
+function htmlToText(html) {
+  let t = String(html || '');
+  t = t.replace(/<!--[\s\S]*?-->/g, ' ');
+  t = t.replace(/<(script|style|noscript|svg|head)[\s\S]*?<\/\1>/gi, ' ');
+  t = t.replace(/<\/(p|div|li|h[1-6]|tr|section|article|br)>/gi, '\n');
+  t = t.replace(/<br\s*\/?>/gi, '\n');
+  t = t.replace(/<[^>]+>/g, ' ');
+  t = decodeEntities(t);
+  t = t.replace(/[ \t\f\v]+/g, ' ');
+  t = t.replace(/\s*\n\s*/g, '\n').replace(/\n{3,}/g, '\n\n');
+  return t.trim();
+}
+
+/* opencode webfetch.ts tarzı HTML→Markdown (hafif sürüm — Turndown bağımlılığı yok):
+   başlıklar, linkler, kalık/italik, kod blokları, listeler korunur */
+function htmlToMarkdown(html) {
+  let h = String(html || '');
+  h = h.replace(/<script[\s\S]*?<\/script>/gi, '').replace(/<style[\s\S]*?<\/style>/gi, '').replace(/<!--[\s\S]*?-->/g, '');
+  h = h.replace(/<pre[^>]*>\s*<code[^>]*>([\s\S]*?)<\/code>\s*<\/pre>/gi, (_m, c) => '\n```\n' + decodeEntities(c) + '\n```\n');
+  h = h.replace(/<code[^>]*>([\s\S]*?)<\/code>/gi, (_m, c) => '`' + decodeEntities(c) + '`');
+  h = h.replace(/<h([1-6])[^>]*>([\s\S]*?)<\/h\1>/gi, (_m, lvl, c) => '\n' + '#'.repeat(Number(lvl)) + ' ' + decodeEntities(c.replace(/<[^>]+>/g, '')).trim() + '\n');
+  h = h.replace(/<a\s+[^>]*href=["']([^"']+)["'][^>]*>([\s\S]*?)<\/a>/gi, (_m, href, c) => `[${decodeEntities(c.replace(/<[^>]+>/g, '')).trim()}](${href})`);
+  h = h.replace(/<(b|strong)[^>]*>([\s\S]*?)<\/\1>/gi, (_m, _t, c) => `**${decodeEntities(c.replace(/<[^>]+>/g, '')).trim()}**`);
+  h = h.replace(/<(i|em)[^>]*>([\s\S]*?)<\/\1>/gi, (_m, _t, c) => `*${decodeEntities(c.replace(/<[^>]+>/g, '')).trim()}*`);
+  h = h.replace(/<li[^>]*>([\s\S]*?)<\/li>/gi, (_m, c) => `\n- ${decodeEntities(c.replace(/<[^>]+>/g, '')).trim()}`);
+  h = h.replace(/<hr\s*\/?>/gi, '\n---\n');
+  h = h.replace(/<br\s*\/?>/gi, '\n');
+  h = h.replace(/<\/(p|div|section|article|tr|h[1-6])>/gi, '\n');
+  h = h.replace(/<[^>]+>/g, '');
+  h = decodeEntities(h);
+  h = h.replace(/[ \t]+/g, ' ').replace(/\n{3,}/g, '\n\n');
+  return h.trim();
+}
+
+/* DuckDuckGo HTML sonuç ayrıştırıcı — test edilebilir saf fonksiyon */
+function parseDdgResults(html, limit = 8) {
+  const out = [];
+  const re =
+    /<a\s+([^>]*\bresult__a\b[^>]*)>([\s\S]*?)<\/a>([\s\S]*?)(?=<a\s+[^>]*\bresult__a\b|$)/g;
+  let m;
+  while ((m = re.exec(html)) && out.length < limit) {
+    const attrs = m[1] || '';
+    const hrefM = attrs.match(/href="([^"]+)"/) || attrs.match(/href='([^']+)'/);
+    let href = hrefM ? hrefM[1] : '';
+    try {
+      const u = new URL(href, 'https://duckduckgo.com');
+      const uddg = u.searchParams.get('uddg');
+      href = uddg ? decodeURIComponent(uddg) : u.toString();
+    } catch {}
+    const title = decodeEntities(m[2] || '').replace(/<[^>]+>/g, '').trim();
+    const seg = m[3] || '';
+    const snM = seg.match(/<a\s+[^>]*result__snippet[^>]*>([\s\S]*?)<\/a>/);
+    const snippet = snM ? decodeEntities(snM[1]).replace(/<[^>]+>/g, '').trim() : '';
+    if (!title || !/^https?:\/\//i.test(href)) continue;
+    if (out.some((r) => r.url === href)) continue;
+    out.push({ title, url: href, snippet });
+  }
+  return out;
+}
+
+async function fetchWithTimeout(url, opts = {}, timeoutMs = 20000, signal) {
+  const ctrl = new AbortController();
+  const timer = setTimeout(() => ctrl.abort(new Error('zaman aşımı')), timeoutMs);
+  const onAbort = () => ctrl.abort(new Error('iptal'));
+  if (signal) {
+    if (signal.aborted) return Promise.reject(new Error('iptal'));
+    signal.addEventListener('abort', onAbort, { once: true });
+  }
+  try {
+    return await fetch(url, { ...opts, signal: ctrl.signal, redirect: 'follow' });
+  } finally {
+    clearTimeout(timer);
+    if (signal) signal.removeEventListener('abort', onAbort);
+  }
+}
+
+async function webSearch(query, { maxResults = 8, signal } = {}) {
+  const q = String(query || '').trim().slice(0, 400);
+  if (!q) return { ok: false, error: 'boş sorgu' };
+  let res;
+  try {
+    res = await fetchWithTimeout(
+      'https://html.duckduckgo.com/html/',
+      {
+        method: 'POST',
+        headers: {
+          'Content-Type': 'application/x-www-form-urlencoded',
+          'User-Agent': UA,
+          Accept: 'text/html',
+        },
+        body: new URLSearchParams({ q }).toString(),
+      },
+      20000,
+      signal
+    );
+  } catch (e) {
+    /* ağ hatası/rate-limit exception fırlatmasın — zincir devam edebilsin */
+    return { ok: false, error: 'DDG erişilemedi: ' + String((e && e.message) || e) };
+  }
+  if (!res.ok) return { ok: false, error: `DDG HTTP ${res.status}` };
+  const html = await res.text();
+  const results = parseDdgResults(html, Math.min(Math.max(Number(maxResults) || 8, 1), 12));
+  if (!results.length) {
+    return { ok: true, query: q, results, note: 'sonuç yok ya da DDG yanıt biçimi değişti' };
+  }
+  return { ok: true, query: q, count: results.length, results };
+}
+
+const MAX_FETCH_CHARS = 9000;
+
+async function httpFetch(url, { maxChars = MAX_FETCH_CHARS, format = 'text', timeoutMs = 30000, signal } = {}) {
+  const safe = assertPublicHttpUrl(url);
+  const t = Math.min(Math.max(Number(timeoutMs) || 30000, 1000), 120000); // opencode: default 30s, max 120s
+  const res = await fetchWithTimeout(
+    safe,
+    { headers: { 'User-Agent': UA, Accept: 'text/html,text/plain,application/json;q=0.9,*/*;q=0.5' } },
+    t,
+    signal
+  );
+  const ctype = String(res.headers.get('content-type') || '');
+  if (!res.ok) throw new Error(`HTTP ${res.status} ${res.statusText}`);
+  const isTexty =
+    /text\/|json|xml|javascript/i.test(ctype) || ctype === '';
+  if (!isTexty) {
+    return { ok: true, url: safe, status: res.status, contentType: ctype, note: 'ikili/metin olmayan içerik indirilmedi' };
+  }
+  const cap = 400000;
+  let body = await res.text();
+  if (body.length > cap) body = body.slice(0, cap);
+  let content;
+  if (/html/i.test(ctype)) {
+    if (format === 'html') content = body;
+    else if (format === 'markdown') content = htmlToMarkdown(body);
+    else content = htmlToText(body);
+  } else {
+    content = body;
+  }
+  const truncated = content.length > maxChars;
+  return {
+    ok: true,
+    url: safe,
+    status: res.status,
+    contentType: ctype,
+    format,
+    truncated,
+    content: content.slice(0, Math.max(1000, Math.min(Number(maxChars) || MAX_FETCH_CHARS, 50000))),
+  };
+}
+
+/* ================= opencode edit.ts BİREBİR PORT =================
+   Kaynak: opencode-dev/packages/opencode/src/tool/edit.ts
+   9 aşamalı replacer zinciri: Simple → LineTrimmed → BlockAnchor →
+   WhitespaceNormalized → IndentationFlexible → EscapeNormalized →
+   TrimmedBoundary → ContextAware → MultiOccurrence.
+   Modelin old_string'i ufak girinti/boşluk farkıyla tutturamadığında zincir
+   akıllı eşleşme bulur — "dosyayı baştan oku" döngüsü kökten kırılır. */
+
+function normalizeLineEndings(text) {
+  return text.replaceAll('\r\n', '\n');
+}
+function detectLineEnding(text) {
+  return text.includes('\r\n') ? '\r\n' : '\n';
+}
+function convertToLineEnding(text, ending) {
+  if (ending === '\n') return text;
+  return text.replaceAll('\n', '\r\n');
+}
+
+/* blok-anchor fallback benzerlik eşikleri (edit.ts:220-221) */
+const SINGLE_CANDIDATE_SIMILARITY_THRESHOLD = 0.65;
+const MULTIPLE_CANDIDATES_SIMILARITY_THRESHOLD = 0.65;
+
+function levenshtein(a, b) {
+  if (a === '' || b === '') return Math.max(a.length, b.length);
+  const matrix = Array.from({ length: a.length + 1 }, (_, i) =>
+    Array.from({ length: b.length + 1 }, (_, j) => (i === 0 ? j : j === 0 ? i : 0))
+  );
+  for (let i = 1; i <= a.length; i++) {
+    for (let j = 1; j <= b.length; j++) {
+      const cost = a[i - 1] === b[j - 1] ? 0 : 1;
+      matrix[i][j] = Math.min(matrix[i - 1][j] + 1, matrix[i][j - 1] + 1, matrix[i - 1][j - 1] + cost);
+    }
+  }
+  return matrix[a.length][b.length];
+}
+
+const SimpleReplacer = function* (_content, find) {
+  yield find;
+};
+
+const LineTrimmedReplacer = function* (content, find) {
+  const originalLines = content.split('\n');
+  const searchLines = find.split('\n');
+  if (searchLines[searchLines.length - 1] === '') searchLines.pop();
+  for (let i = 0; i <= originalLines.length - searchLines.length; i++) {
+    let matches = true;
+    for (let j = 0; j < searchLines.length; j++) {
+      if (originalLines[i + j].trim() !== searchLines[j].trim()) {
+        matches = false;
+        break;
+      }
+    }
+    if (matches) {
+      let matchStartIndex = 0;
+      for (let k = 0; k < i; k++) matchStartIndex += originalLines[k].length + 1;
+      let matchEndIndex = matchStartIndex;
+      for (let k = 0; k < searchLines.length; k++) {
+        matchEndIndex += originalLines[i + k].length;
+        if (k < searchLines.length - 1) matchEndIndex += 1;
+      }
+      yield content.substring(matchStartIndex, matchEndIndex);
+    }
+  }
+};
+
+const BlockAnchorReplacer = function* (content, find) {
+  const originalLines = content.split('\n');
+  const searchLines = find.split('\n');
+  if (searchLines.length < 3) return;
+  if (searchLines[searchLines.length - 1] === '') searchLines.pop();
+  const firstLineSearch = searchLines[0].trim();
+  const lastLineSearch = searchLines[searchLines.length - 1].trim();
+  const searchBlockSize = searchLines.length;
+  const maxLineDelta = Math.max(1, Math.floor(searchBlockSize * 0.25));
+  const candidates = [];
+  for (let i = 0; i < originalLines.length; i++) {
+    if (originalLines[i].trim() !== firstLineSearch) continue;
+    for (let j = i + 2; j < originalLines.length; j++) {
+      if (originalLines[j].trim() === lastLineSearch) {
+        const actualBlockSize = j - i + 1;
+        if (Math.abs(actualBlockSize - searchBlockSize) <= maxLineDelta) {
+          candidates.push({ startLine: i, endLine: j });
+        }
+        break;
+      }
+    }
+  }
+  if (candidates.length === 0) return;
+  if (candidates.length === 1) {
+    const { startLine, endLine } = candidates[0];
+    const actualBlockSize = endLine - startLine + 1;
+    let similarity = 0;
+    const linesToCheck = Math.min(searchBlockSize - 2, actualBlockSize - 2);
+    if (linesToCheck > 0) {
+      for (let j = 1; j < searchBlockSize - 1 && j < actualBlockSize - 1; j++) {
+        const originalLine = originalLines[startLine + j].trim();
+        const searchLine = searchLines[j].trim();
+        const maxLen = Math.max(originalLine.length, searchLine.length);
+        if (maxLen === 0) continue;
+        const distance = levenshtein(originalLine, searchLine);
+        similarity += (1 - distance / maxLen) / linesToCheck;
+        if (similarity >= SINGLE_CANDIDATE_SIMILARITY_THRESHOLD) break;
+      }
+    } else {
+      similarity = 1.0;
+    }
+    if (similarity >= SINGLE_CANDIDATE_SIMILARITY_THRESHOLD) {
+      let matchStartIndex = 0;
+      for (let k = 0; k < startLine; k++) matchStartIndex += originalLines[k].length + 1;
+      let matchEndIndex = matchStartIndex;
+      for (let k = startLine; k <= endLine; k++) {
+        matchEndIndex += originalLines[k].length;
+        if (k < endLine) matchEndIndex += 1;
+      }
+      yield content.substring(matchStartIndex, matchEndIndex);
+    }
+    return;
+  }
+  let bestMatch = null;
+  let maxSimilarity = -1;
+  for (const candidate of candidates) {
+    const { startLine, endLine } = candidate;
+    const actualBlockSize = endLine - startLine + 1;
+    let similarity = 0;
+    const linesToCheck = Math.min(searchBlockSize - 2, actualBlockSize - 2);
+    if (linesToCheck > 0) {
+      for (let j = 1; j < searchBlockSize - 1 && j < actualBlockSize - 1; j++) {
+        const originalLine = originalLines[startLine + j].trim();
+        const searchLine = searchLines[j].trim();
+        const maxLen = Math.max(originalLine.length, searchLine.length);
+        if (maxLen === 0) continue;
+        const distance = levenshtein(originalLine, searchLine);
+        similarity += 1 - distance / maxLen;
+      }
+      similarity /= linesToCheck;
+    } else {
+      similarity = 1.0;
+    }
+    if (similarity > maxSimilarity) {
+      maxSimilarity = similarity;
+      bestMatch = candidate;
+    }
+  }
+  if (maxSimilarity >= MULTIPLE_CANDIDATES_SIMILARITY_THRESHOLD && bestMatch) {
+    const { startLine, endLine } = bestMatch;
+    let matchStartIndex = 0;
+    for (let k = 0; k < startLine; k++) matchStartIndex += originalLines[k].length + 1;
+    let matchEndIndex = matchStartIndex;
+    for (let k = startLine; k <= endLine; k++) {
+      matchEndIndex += originalLines[k].length;
+      if (k < endLine) matchEndIndex += 1;
+    }
+    yield content.substring(matchStartIndex, matchEndIndex);
+  }
+};
+
+const WhitespaceNormalizedReplacer = function* (content, find) {
+  const normalizeWhitespace = (text) => text.replace(/\s+/g, ' ').trim();
+  const normalizedFind = normalizeWhitespace(find);
+  const lines = content.split('\n');
+  for (let i = 0; i < lines.length; i++) {
+    const line = lines[i];
+    if (normalizeWhitespace(line) === normalizedFind) {
+      yield line;
+    } else {
+      const normalizedLine = normalizeWhitespace(line);
+      if (normalizedLine.includes(normalizedFind)) {
+        const words = find.trim().split(/\s+/);
+        if (words.length > 0) {
+          const pattern = words.map((word) => word.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')).join('\\s+');
+          try {
+            const regex = new RegExp(pattern);
+            const match = line.match(regex);
+            if (match) yield match[0];
+          } catch {}
+        }
+      }
+    }
+  }
+  const findLines = find.split('\n');
+  if (findLines.length > 1) {
+    for (let i = 0; i <= lines.length - findLines.length; i++) {
+      const block = lines.slice(i, i + findLines.length);
+      if (normalizeWhitespace(block.join('\n')) === normalizedFind) {
+        yield block.join('\n');
+      }
+    }
+  }
+};
+
+const IndentationFlexibleReplacer = function* (content, find) {
+  const removeIndentation = (text) => {
+    const lines = text.split('\n');
+    const nonEmptyLines = lines.filter((line) => line.trim().length > 0);
+    if (nonEmptyLines.length === 0) return text;
+    const minIndent = Math.min(
+      ...nonEmptyLines.map((line) => {
+        const match = line.match(/^(\s*)/);
+        return match ? match[1].length : 0;
+      })
+    );
+    return lines.map((line) => (line.trim().length === 0 ? line : line.slice(minIndent))).join('\n');
+  };
+  const normalizedFind = removeIndentation(find);
+  const contentLines = content.split('\n');
+  const findLines = find.split('\n');
+  for (let i = 0; i <= contentLines.length - findLines.length; i++) {
+    const block = contentLines.slice(i, i + findLines.length).join('\n');
+    if (removeIndentation(block) === normalizedFind) yield block;
+  }
+};
+
+const EscapeNormalizedReplacer = function* (content, find) {
+  const unescapeString = (str) =>
+    str.replace(/\\(n|t|r|'|"|`|\\|\n|\$)/g, (match, capturedChar) => {
+      switch (capturedChar) {
+        case 'n': return '\n';
+        case 't': return '\t';
+        case 'r': return '\r';
+        case "'": return "'";
+        case '"': return '"';
+        case '`': return '`';
+        case '\\': return '\\';
+        case '\n': return '\n';
+        case '$': return '$';
+        default: return match;
+      }
+    });
+  const unescapedFind = unescapeString(find);
+  if (content.includes(unescapedFind)) yield unescapedFind;
+  const lines = content.split('\n');
+  const findLines = unescapedFind.split('\n');
+  for (let i = 0; i <= lines.length - findLines.length; i++) {
+    const block = lines.slice(i, i + findLines.length).join('\n');
+    if (unescapeString(block) === unescapedFind) yield block;
+  }
+};
+
+const MultiOccurrenceReplacer = function* (content, find) {
+  let startIndex = 0;
+  while (true) {
+    const index = content.indexOf(find, startIndex);
+    if (index === -1) break;
+    yield find;
+    startIndex = index + find.length;
+  }
+};
+
+const TrimmedBoundaryReplacer = function* (content, find) {
+  const trimmedFind = find.trim();
+  if (trimmedFind === find) return;
+  if (content.includes(trimmedFind)) yield trimmedFind;
+  const lines = content.split('\n');
+  const findLines = find.split('\n');
+  for (let i = 0; i <= lines.length - findLines.length; i++) {
+    const block = lines.slice(i, i + findLines.length).join('\n');
+    if (block.trim() === trimmedFind) yield block;
+  }
+};
+
+const ContextAwareReplacer = function* (content, find) {
+  const findLines = find.split('\n');
+  if (findLines.length < 3) return;
+  if (findLines[findLines.length - 1] === '') findLines.pop();
+  const contentLines = content.split('\n');
+  const firstLine = findLines[0].trim();
+  const lastLine = findLines[findLines.length - 1].trim();
+  for (let i = 0; i < contentLines.length; i++) {
+    if (contentLines[i].trim() !== firstLine) continue;
+    for (let j = i + 2; j < contentLines.length; j++) {
+      if (contentLines[j].trim() === lastLine) {
+        const blockLines = contentLines.slice(i, j + 1);
+        const block = blockLines.join('\n');
+        if (blockLines.length === findLines.length) {
+          let matchingLines = 0;
+          let totalNonEmptyLines = 0;
+          for (let k = 1; k < blockLines.length - 1; k++) {
+            const blockLine = blockLines[k].trim();
+            const findLine = findLines[k].trim();
+            if (blockLine.length > 0 || findLine.length > 0) {
+              totalNonEmptyLines++;
+              if (blockLine === findLine) matchingLines++;
+            }
+          }
+          if (totalNonEmptyLines === 0 || matchingLines / totalNonEmptyLines >= 0.5) {
+            yield block;
+            break;
+          }
+        }
+        break;
+      }
+    }
+  }
+};
+
+function isDisproportionateMatch(search, oldString) {
+  const oldLines = oldString.split('\n').length;
+  const searchLines = search.split('\n').length;
+  if (searchLines >= Math.max(oldLines + 3, oldLines * 2)) return true;
+  if (oldLines === 1) return false;
+  return search.trim().length > Math.max(oldString.trim().length + 500, oldString.trim().length * 4);
+}
+
+/* edit.ts:682-729 replace() — zinciri sırayla dener; tek eşleşme şart,
+   replaceAll'de tümünü değiştirir; hata mesajları BİREBİR opencode */
+function ocReplace(content, oldString, newString, replaceAll = false) {
+  if (oldString === newString) {
+    throw new Error('No changes to apply: oldString and newString are identical.');
+  }
+  if (oldString === '') {
+    throw new Error(
+      'oldString cannot be empty when editing an existing file. Provide the exact text to replace, or use write_file for an intentional full-file replacement.'
+    );
+  }
+  let notFound = true;
+  for (const replacer of [
+    SimpleReplacer,
+    LineTrimmedReplacer,
+    BlockAnchorReplacer,
+    WhitespaceNormalizedReplacer,
+    IndentationFlexibleReplacer,
+    EscapeNormalizedReplacer,
+    TrimmedBoundaryReplacer,
+    ContextAwareReplacer,
+    MultiOccurrenceReplacer,
+  ]) {
+    for (const search of replacer(content, oldString)) {
+      const index = content.indexOf(search);
+      if (index === -1) continue;
+      notFound = false;
+      if (isDisproportionateMatch(search, oldString)) {
+        throw new Error(
+          'Refusing replacement because the matched span is much larger than oldString. Re-read the file and provide the full exact oldString for the intended replacement.'
+        );
+      }
+      if (replaceAll) {
+        return { result: content.replaceAll(search, newString), replacements: content.split(search).length - 1 };
+      }
+      const lastIndex = content.lastIndexOf(search);
+      if (index !== lastIndex) continue;
+      return { result: content.substring(0, index) + newString + content.substring(index + search.length), replacements: 1 };
+    }
+  }
+  if (notFound) {
+    throw new Error(
+      'Could not find oldString in the file. It must match exactly, including whitespace, indentation, and line endings.'
+    );
+  }
+  throw new Error('Found multiple matches for oldString. Provide more surrounding context to make the match unique.');
+}
+
+/* ---------- opencode diff portu (npm "diff" paketinin diffLines karşılığı) ----------
+   LCS tabanlı satır diff'i: additions/deletions sayımı (edit.ts:175-180) ve
+   UI split-diff görünümü için kırpılmış bölge (trimDiff mantığı). */
+function diffLineCounts(aText, bText) {
+  /* bos metin = 0 satir (opencode npm diffLines davranisi) */
+  const a = aText ? String(aText).split('\n') : [];
+  const b = bText ? String(bText).split('\n') : [];
+  let p = 0;
+  while (p < a.length && p < b.length && a[p] === b[p]) p++;
+  let ea = a.length - 1;
+  let eb = b.length - 1;
+  while (ea >= p && eb >= p && a[ea] === b[eb]) { ea--; eb--; }
+  const midA = a.slice(p, ea + 1);
+  const midB = b.slice(p, eb + 1);
+  const n = midA.length;
+  const m = midB.length;
+  if (!n && !m) return { additions: 0, deletions: 0 };
+  if (n * m > 400000 || n > 1500 || m > 1500) return { additions: m, deletions: n };
+  const w = m + 1;
+  const dp = new Int32Array((n + 1) * w);
+  for (let i = n - 1; i >= 0; i--) {
+    for (let j = m - 1; j >= 0; j--) {
+      dp[i * w + j] =
+        midA[i] === midB[j]
+          ? dp[(i + 1) * w + j + 1] + 1
+          : Math.max(dp[(i + 1) * w + j], dp[i * w + j + 1]);
+    }
+  }
+  const matched = dp[0];
+  return { additions: m - matched, deletions: n - matched };
+}
+
+const DIFF_REGION_CTX = 3; /* değişiklik etrafında kaç bağlam satırı gösterilir */
+const DIFF_REGION_CAP = 3500; /* UI'ye giden bölge başına karakter tavanı */
+
+function capDiffText(s) {
+  const t = String(s || '');
+  return t.length > DIFF_REGION_CAP ? t.slice(0, DIFF_REGION_CAP) + '\n…' : t;
+}
+
+/* Değişen bölgeyi ±DIFF_REGION_CTX bağlam satırıyla kırpar (opencode trimDiff
+   mantığı — tüm dosya yerine yalnız değişen kısım UI'ye gider). */
+function diffRegion(before, after) {
+  const a = String(before || '').split('\n');
+  const b = String(after || '').split('\n');
+  let p = 0;
+  while (p < a.length && p < b.length && a[p] === b[p]) p++;
+  let ea = a.length - 1;
+  let eb = b.length - 1;
+  while (ea >= p && eb >= p && a[ea] === b[eb]) { ea--; eb--; }
+  const start = Math.max(0, p - DIFF_REGION_CTX);
+  const beforeRegion = a.slice(start, Math.min(a.length, ea + 1 + DIFF_REGION_CTX)).join('\n');
+  const afterRegion = b.slice(start, Math.min(b.length, eb + 1 + DIFF_REGION_CTX)).join('\n');
+  return { before: capDiffText(beforeRegion), after: capDiffText(afterRegion), startLine: start + 1 };
+}
+
+/* ---------- read_file disk-cache (opencode file-state portu) ----------
+   mtime+size değişmemişse dosya yeniden diskten OKUNMAZ — cache'ten döner.
+   edit_file/write_file kendi yazdıklarından sonra cache'i düşürür; böylece
+   sonraki okuma daima güncel içerik verir. */
+const _readCache = new Map(); /* abs → { mtimeMs, size, raw } */
+
+function readCacheGet(abs, st) {
+  const hit = _readCache.get(abs);
+  if (hit && hit.mtimeMs === st.mtimeMs && hit.size === st.size) return hit.raw;
+  const raw = fs.readFileSync(abs, 'utf8');
+  if (_readCache.size > 200) _readCache.delete(_readCache.keys().next().value);
+  _readCache.set(abs, { mtimeMs: st.mtimeMs, size: st.size, raw });
+  return raw;
+}
+
+function readCacheDrop(abs) {
+  _readCache.delete(abs);
+}
+
+const definitions = [
+  {
+    type: 'function',
+    function: {
+      name: 'run_command',
+      description:
+        'Executes a given PowerShell command on the user\'s Windows machine in the workspace directory and returns combined stdout/stderr. Use this tool for terminal operations like builds, git, npm installs, docker, etc. DO NOT use it for file operations (reading, writing, editing, searching, finding files) - use the specialized tools for this instead: read_file, write_file, edit_file, grep, glob.',
+      parameters: {
+        type: 'object',
+        properties: {
+          command: { type: 'string', description: 'PowerShell command line to execute' },
+          timeout_ms: { type: 'number', description: 'Optional timeout in ms (default 120000)' },
+        },
+        required: ['command'],
+      },
+    },
+  },
+  {
+    type: 'function',
+    function: {
+      name: 'read_file',
+      /* opencode read.txt BİREBİR port (parametre adı path olarak kaldı) */
+      description:
+        'Read a file or directory from the local filesystem. If the path does not exist, an error is returned.\n\n' +
+        'Usage:\n' +
+        '- By default, this tool returns up to 2000 lines from the start of the file.\n' +
+        '- The offset parameter is the line number to start reading from (1-indexed).\n' +
+        '- To read later sections, call this tool again with a larger offset — NEVER re-read from the start of the file.\n' +
+        '- Use the grep tool to find specific content in large files or files with long lines.\n' +
+        '- If you are unsure of the correct file path, use the glob tool to look up filenames by glob pattern.\n' +
+        '- Contents are returned with each line prefixed by its line number as `<line>: <content>`. For example, if a file has contents "foo\\n", you will receive "1: foo\\n".\n' +
+        '- Any line longer than 2000 characters is truncated.\n' +
+        '- Call this tool in parallel when you know there are multiple files you want to read.\n' +
+        '- Avoid tiny repeated slices (30 line chunks). If you need more context, read a larger window.\n' +
+        '- File contents you read STAY in your context for the WHOLE session. Do NOT re-read a file you already read: chain edits from your previous read plus your own edit_file/write_file results. Re-reading the same file (same or overlapping range) wastes turns and is forbidden unless the file actually changed on disk outside your own edits.',
+      parameters: {
+        type: 'object',
+        properties: {
+          path: { type: 'string', description: 'The path to the file to read' },
+          offset: { type: 'number', description: 'The line number to start reading from (1-indexed)' },
+          limit: { type: 'number', description: 'The maximum number of lines to read (defaults to 2000)' },
+        },
+        required: ['path'],
+      },
+    },
+  },
+  {
+    type: 'function',
+    function: {
+      name: 'write_file',
+      /* opencode write.txt BİREBİR port */
+      description:
+        'Writes a file to the local filesystem.\n\n' +
+        'Usage:\n' +
+        '- This tool will overwrite the existing file if there is one at the provided path.\n' +
+        '- If this is an existing file, you MUST use the read_file tool first to read the file\'s contents. This tool will fail if you did not read the file first.\n' +
+        '- ALWAYS prefer editing existing files in the codebase with edit_file. NEVER write new files unless explicitly required.\n' +
+        '- The result includes additions/deletions counts — the change is APPLIED to disk immediately; do NOT read the file again to verify.',
+      parameters: {
+        type: 'object',
+        properties: {
+          path: { type: 'string', description: 'The path to the file to write' },
+          content: { type: 'string', description: 'The content to write to the file' },
+        },
+        required: ['path', 'content'],
+      },
+    },
+  },
+  {
+    type: 'function',
+    function: {
+      name: 'edit_file',
+      /* opencode edit.txt BİREBİR port (parametre adları snake_case kaldı) */
+      description:
+        'Performs exact string replacements in files.\n\n' +
+        'Usage:\n' +
+        '- You must use your read_file tool at least once in the conversation before editing. This tool will error if you attempt an edit without reading the file.\n' +
+        '- When editing text from read_file output, ensure you preserve the exact indentation (tabs/spaces) as it appears AFTER the line number prefix. The line number prefix format is: line number + colon + space (e.g., `1: `). Everything after that space is the actual file content to match. Never include any part of the line number prefix in the old_string or new_string.\n' +
+        '- ALWAYS prefer editing existing files in the codebase. NEVER write new files unless explicitly required.\n' +
+        '- The edit will FAIL if `old_string` is not found in the file with an error "oldString not found in content".\n' +
+        '- The edit will FAIL if `old_string` is found multiple times in the file with an error "Found multiple matches for oldString. Provide more surrounding lines in old_string to identify the correct match." Either provide a larger string with more surrounding context to make it unique or use `replace_all` to change every instance of `old_string`.\n' +
+        '- Use `replace_all` for replacing and renaming strings across the file. This parameter is useful if you want to rename a variable for instance.\n' +
+        '- The result includes additions/deletions counts — the change is APPLIED to disk immediately; do NOT read the file again to verify.\n' +
+        '- Chain follow-up edits from what you already read plus your own previous edits — do NOT re-read the file between edits. Multiple edit_file calls to different files can be issued in PARALLEL in one turn.',
+      parameters: {
+        type: 'object',
+        properties: {
+          path: { type: 'string', description: 'The path to the file to modify' },
+          old_string: { type: 'string', description: 'The text to replace' },
+          new_string: { type: 'string', description: 'The text to replace it with (must be different from old_string)' },
+          replace_all: { type: 'boolean', description: 'Replace all occurrences of old_string (default false)' },
+        },
+        required: ['path', 'old_string', 'new_string'],
+      },
+    },
+  },
+  {
+    type: 'function',
+    function: {
+      name: 'grep',
+      description:
+        'Fast content search tool that works with any codebase size. Searches file contents using regular expressions (case-SENSITIVE by default — pass case_insensitive:true to ignore case); supports full regex syntax (eg. "log.*Error", "function\\s+\\w+"). Filter files by pattern with include (eg. "*.js", "*.{ts,tsx}"). Returns grouped matches as `<path>:` + `  Line N: text`, capped at 100 matches. node_modules/.git and similar dirs are skipped. Use this to find where functions/symbols/errors live before editing.',
+      parameters: {
+        type: 'object',
+        properties: {
+          pattern: { type: 'string', description: 'Regular expression' },
+          path: { type: 'string', description: 'File or directory to search; defaults to workspace root' },
+          include: { type: 'string', description: 'Glob filter like "*.js" or "*.{ts,tsx}"' },
+          case_insensitive: { type: 'boolean', description: 'Ignore case (default false — search is case-sensitive)' },
+        },
+        required: ['pattern'],
+      },
+    },
+  },
+  {
+    type: 'function',
+    function: {
+      name: 'glob',
+      description:
+        'Fast file pattern matching tool that works with any codebase size. Supports glob patterns like "**/*.js" or "src/**/*.ts". Returns matching file paths (max 100; more specific pattern/path if truncated). Use when you need to find files by name patterns; batch multiple speculative searches in one turn.',
+      parameters: {
+        type: 'object',
+        properties: {
+          pattern: { type: 'string', description: 'Glob pattern, e.g. "src/**/*.ts"' },
+          path: { type: 'string', description: 'Directory to search; defaults to workspace root' },
+        },
+        required: ['pattern'],
+      },
+    },
+  },
+  {
+    type: 'function',
+    function: {
+      name: 'list_dir',
+      description: 'List entries of a directory (localeCompare sorted, directories suffixed with `/`, max 500 entries) with sizes and types.',
+      parameters: {
+        type: 'object',
+        properties: {
+          path: { type: 'string', description: 'Defaults to workspace root' },
+        },
+        required: [],
+      },
+    },
+  },
+  {
+    type: 'function',
+    function: {
+      name: 'webfetch',
+      description:
+        '- Fetches content from a specified URL\n' +
+        '- Takes a URL and optional format as input\n' +
+        '- Fetches the URL content, converts to requested format (markdown by default)\n' +
+        '- Returns the content in the specified format\n' +
+        '- Use this tool when you need to retrieve and analyze web content\n' +
+        '- IMPORTANT: if another tool is present that offers better web fetching capabilities, is more targeted to the task, or has fewer restrictions, prefer using that tool instead of this one.\n' +
+        '- The URL must be a fully-formed valid URL\n' +
+        '- HTTP URLs will be automatically upgraded to HTTPS\n' +
+        '- Format options: "text" (default), "markdown", or "html"\n' +
+        '- Local/private network addresses are blocked; timeout default 30s (max 120s)',
+      parameters: {
+        type: 'object',
+        properties: {
+          url: { type: 'string', format: 'uri', description: 'The URL to fetch content from' },
+          format: { type: 'string', enum: ['text', 'markdown', 'html'], description: 'The format to return the content in (defaults to text)' },
+          timeout: { type: 'number', description: 'Optional timeout in seconds (max 120)' },
+          max_chars: { type: 'number', description: 'Output character cap (default 9000, max 50000)' },
+        },
+        required: ['url'],
+      },
+    },
+  },
+  {
+    type: 'function',
+    function: {
+      name: 'web_search',
+      description:
+        'FAST web search with an automatic chain (order is configurable in Ayarlar → Web Arama): local SearXNG (free, unlimited, auto-prioritized when running — start with `beast searxng`), built-in browser (real Chromium searching GOOGLE directly with AI Mode — no bot protection; the response may include an `ai` field holding Google\'s own AI answer), TinyFish API (only if a key is configured), then Python multi-engine (ddgs / DuckDuckGo+Bing+Mojeek). If the browser hits CAPTCHA/unusual traffic it is skipped for 10 minutes and the next engine takes over. Returns {ai?, results[{title,url,snippet}]}. Use when fresh or external info is needed; skip for things you already know.',
+      parameters: {
+        type: 'object',
+        properties: {
+          query: { type: 'string' },
+          max_results: { type: 'number', description: '1-12, default 8' },
+        },
+        required: ['query'],
+      },
+    },
+  },
+  {
+    type: 'function',
+    function: {
+      name: 'http_fetch',
+      description:
+        'Fetch one URL and return its text content (HTML converted to plain text, JSON as-is). Local/private network addresses are blocked.',
+      parameters: {
+        type: 'object',
+        properties: {
+          url: { type: 'string', format: 'uri' },
+          max_chars: { type: 'number', description: 'default 9000' },
+        },
+        required: ['url'],
+      },
+    },
+  },
+  {
+    type: 'function',
+    function: {
+      name: 'deep_search',
+      description:
+        'AGENTIC DEEP RESEARCH — use when web_search is not enough or the answer was NOT found on the first try. Runs 1-4 query variants IN PARALLEL (rephrase, synonyms, Turkish + English spellings of the same question), merges + dedupes results, then AUTOMATICALLY opens and reads the top pages with a HIDDEN real-Chromium browser (JS/SPA pages work; the visible panel is NOT touched) and returns full-text excerpts. Ideal for: multi-angle questions (price comparison, reviews, specs, "find everything about X"), Turkish queries that miss results, pages that need JS rendering. Returns {queries, results[{title,url,snippet}], pages[{url,title,content}]}.',
+      parameters: {
+        type: 'object',
+        properties: {
+          queries: {
+            type: 'array',
+            maxItems: 4,
+            items: { type: 'string' },
+            description: '1-4 query variants; e.g. ["iphone 16 fiyat", "iphone 16 price turkey", "iphone 16 technosa"]',
+          },
+          max_results: { type: 'number', description: 'merged result cap, default 16' },
+          read_top: { type: 'number', description: 'how many top results to auto-read fully (0-6, default 3); 0 = results only' },
+        },
+        required: ['queries'],
+      },
+    },
+  },
+  {
+    type: 'function',
+    function: {
+      name: 'python_run',
+      description:
+        "Run Python (system interpreter, or Beast auto-installs a portable embedded runtime on first use). Either pass inline `code`, or `script` = filename inside the Beast scripts library (%APPDATA%\\beast\\scripts — e.g. news.py haber toplayıcı) or an absolute path. Use for scraping, RSS/haber toplama, veri analizi, regex/parsing işleri. stdout+stderr returned; scripts should print results.",
+      parameters: {
+        type: 'object',
+        properties: {
+          code: { type: 'string', description: 'Inline Python source (use this OR script)' },
+          script: { type: 'string', description: 'Script name in beast scripts dir or absolute path' },
+          args: {
+            type: 'array',
+            items: { type: 'string' },
+            description: 'CLI arguments passed to the script, e.g. ["--limit","5","--json"]',
+          },
+          timeout_ms: { type: 'number', description: 'default 120000' },
+        },
+      },
+    },
+  },
+];
+
+async function exec(name, args, ctx) {
+  const cwd = ctx.cwd;
+  try {
+    switch (name) {
+      case 'run_command': {
+        /* opencode shell.ts kuralları: default 120s, negatif timeout reddi */
+        const tRaw = Number(args.timeout_ms);
+        if (Number.isFinite(tRaw) && args.timeout_ms != null && tRaw <= 0) {
+          return JSON.stringify({ ok: false, error: `Invalid timeout value: ${tRaw}. Timeout must be a positive number.` });
+        }
+        const t = Number.isFinite(tRaw) && tRaw > 0 ? tRaw : 120000;
+        const shell = String(args.shell || 'powershell').toLowerCase();
+        const r =
+          shell === 'bash' || shell === 'sh'
+            ? await runBashCommand(String(args.command || ''), cwd, t, ctx.signal)
+            : await runShellCommand(String(args.command || ''), cwd, t, ctx.signal);
+        const b = boundToolOutput((r && r.output) || '');
+        return JSON.stringify({
+          ok: !!(r && r.ok),
+          code: r && r.code,
+          output: b.text || '(no output)',
+          ...(b.outputFile ? { outputFile: b.outputFile } : {}),
+        });
+      }
+      case 'edit_file': {
+        /* opencode edit.ts execute BİREBİR port — replacer zinciri + satır sonu
+           normalizasyonu + diff metadata (additions/deletions + UI diffView) */
+        const filePath = safeResolve(String(args.path || args.filePath || ''), cwd);
+        const oldS = String(args.old_string ?? args.oldString ?? '');
+        const newS = String(args.new_string ?? args.newString ?? '');
+        try {
+          if (oldS === newS) {
+            throw new Error('No changes to apply: oldString and newString are identical.');
+          }
+          /* boş oldString = YENİ dosya oluşturma yolu (edit.ts:90-121) */
+          if (oldS === '') {
+            if (fs.existsSync(filePath)) {
+              throw new Error(
+                'oldString cannot be empty when editing an existing file. Provide the exact text to replace, or use write_file for an intentional full-file replacement.'
+              );
+            }
+            fs.mkdirSync(path.dirname(filePath), { recursive: true });
+            fs.writeFileSync(filePath, newS, 'utf8');
+            readCacheDrop(filePath);
+            const counts = diffLineCounts('', newS);
+            const out = { ok: true, path: filePath, note: 'Edit applied successfully.', replacements: 1, ...counts };
+            if (ctx.wantDiff) out.diffView = { path: filePath, before: '', after: capDiffText(newS), startLine: 1, ...counts };
+            return JSON.stringify(out);
+          }
+          if (!fs.existsSync(filePath)) throw new Error(`File ${filePath} not found`);
+          if (fs.statSync(filePath).isDirectory()) throw new Error(`Path is a directory, not a file: ${filePath}`);
+          const contentOld = fs.readFileSync(filePath, 'utf8');
+          /* satır sonu stili dosyadan alınır, old/new buna çevrilir (edit.ts:129-131) */
+          const ending = detectLineEnding(contentOld);
+          const oldN = convertToLineEnding(normalizeLineEndings(oldS), ending);
+          const newN = convertToLineEnding(normalizeLineEndings(newS), ending);
+          const rep = ocReplace(contentOld, oldN, newN, !!(args.replace_all || args.replaceAll));
+          const contentNew = rep.result;
+          fs.writeFileSync(filePath, contentNew, 'utf8');
+          readCacheDrop(filePath); /* sonraki read_file taze içerik okusun */
+          const counts = diffLineCounts(contentOld, contentNew);
+          const out = {
+            ok: true,
+            path: filePath,
+            note: 'Edit applied successfully.',
+            replacements: rep.replacements,
+            ...counts,
+          };
+          if (ctx.wantDiff) {
+            out.diffView = { path: filePath, ...diffRegion(contentOld, contentNew), ...counts };
+          }
+          return JSON.stringify(out);
+        } catch (e) {
+          return JSON.stringify({ ok: false, error: String((e && e.message) || e) });
+        }
+      }
+      case 'read_file': {
+        /* opencode read.ts BİREBİR port: fuzzy not-found, binary tespiti,
+           dizin okuma, `N: içerik` formatı, 2000 satır / 2000 karakter /
+           50KB tavanları + opencode devam footer'ları */
+        const abs = safeResolve(String(args.path || ''), cwd);
+        if (!fs.existsSync(abs)) {
+          const sibs = fuzzySiblings(abs);
+          const hint = sibs.length ? '\nDid you mean one of these?\n' + sibs.map((s) => `- ${path.join(path.dirname(abs), s)}`).join('\n') : '';
+          return JSON.stringify({ ok: false, error: `File not found: ${abs}${hint}` });
+        }
+        const st = fs.statSync(abs);
+        /* DİZİN OKUMA (opencode read.ts dizin modu): localeCompare sıralı,
+           klasörler `/` ile, offset/limit sayfalı */
+        if (st.isDirectory()) {
+          let names = fs
+            .readdirSync(abs)
+            .sort((a, b) => a.localeCompare(b))
+            .map((name) => {
+              let isDir = false;
+              try { isDir = fs.statSync(path.join(abs, name)).isDirectory(); } catch {}
+              return isDir ? name + '/' : name;
+            });
+          const total = names.length;
+          const offset = Math.max(1, Math.floor(Number(args.offset) || 1));
+          const limit = Math.min(2000, Math.max(1, Math.floor(Number(args.limit) || 2000)));
+          names = names.slice(offset - 1, offset - 1 + limit);
+          const note =
+            names.length < total
+              ? `(Showing ${names.length} of ${total} entries. Use offset parameter to paginate.)`
+              : `(End of directory - total ${total} entries)`;
+          return JSON.stringify({
+            ok: true,
+            path: abs,
+            type: 'directory',
+            totalEntries: total,
+            offset,
+            truncated: offset - 1 + names.length < total,
+            note,
+            content: names.join('\n'),
+          });
+        }
+        if (st.size > MAX_FILE_CHARS * 2) {
+          return JSON.stringify({ ok: false, error: `file too large (${st.size} bytes)` });
+        }
+        /* binary tespiti uzantıdan — PDF'e dokunma (Beast'in pdf hattı var) */
+        if (BINARY_EXT_RE.test(abs)) {
+          return JSON.stringify({ ok: false, error: `Cannot read binary file: ${abs}` });
+        }
+        /* PDF: pdf-parse varsa metin çıkar (v2 class API — bkz. src/agent/pdf.js) */
+        if (/\.pdf$/i.test(abs)) {
+          let extract = null;
+          try { extract = require('./pdf').extract; } catch {}
+          if (!extract) return JSON.stringify({ ok: false, error: 'pdf okuma yok — npm i pdf-parse' });
+          try {
+            const r = await extract(fs.readFileSync(abs));
+            let content = String((r && r.text) || '').trim();
+            const pages = (r && r.total) || '?';
+            const truncated = content.length > MAX_FILE_CHARS;
+            if (truncated) content = content.slice(0, MAX_FILE_CHARS);
+            if (!content.trim()) return JSON.stringify({ ok: false, error: 'pdf metni çıkarılamadı (taranmış görsel olabilir)' });
+            return JSON.stringify({ ok: true, path: abs, pages, truncated, content });
+          } catch (e2) {
+            return JSON.stringify({ ok: false, error: 'pdf okuma hata: ' + String((e2 && e2.message) || e2) });
+          }
+        }
+        /* binary tespiti: ilk 4KB'ta NUL / %30+ yazdırılamaz karakter (opencode kuralı) */
+        try {
+          const fd = fs.openSync(abs, 'r');
+          let probe;
+          try {
+            probe = Buffer.alloc(Math.min(4096, st.size));
+            fs.readSync(fd, probe, 0, probe.length, 0);
+          } finally {
+            fs.closeSync(fd);
+          }
+          if (looksBinary(probe)) {
+            return JSON.stringify({ ok: false, error: `Cannot read binary file: ${abs}` });
+          }
+        } catch (eProbe) {
+          if (eProbe && /Cannot read binary/.test(String(eProbe.message || ''))) throw eProbe;
+        }
+        let raw;
+        try {
+          raw = readCacheGet(abs, st);
+        } catch {
+          raw = fs.readFileSync(abs, 'utf8');
+        }
+        /* opencode read.ts port: satır numaralı çıktı (`N: içerik`), offset/limit
+           penceresi (1-indexed), 2000 satır varsayılan, uzun satır kırpma */
+        const allLines = raw.split('\n');
+        const offset = Math.max(1, Math.floor(Number(args.offset) || 1));
+        const limit = Math.min(2000, Math.max(1, Math.floor(Number(args.limit) || 2000)));
+        let slice = allLines
+          .slice(offset - 1, offset - 1 + limit)
+          .map((l, i) => {
+            const line = l.length > 2000 ? l.slice(0, 2000) + '... (line truncated to 2000 chars)' : l;
+            return `${offset + i}: ${line}`;
+          });
+        let truncated = offset - 1 + limit < allLines.length;
+        let content = slice.join('\n');
+        /* opencode 50KB byte tavanı: aşarsa satır bazında kes + özel footer */
+        const MAX_BYTES = 50 * 1024;
+        let byteCapped = false;
+        if (Buffer.byteLength(content, 'utf8') > MAX_BYTES) {
+          const keep = [];
+          let used = 0;
+          for (const l of slice) {
+            const need = Buffer.byteLength(l, 'utf8') + 1;
+            if (used + need > MAX_BYTES) break;
+            keep.push(l);
+            used += need;
+          }
+          slice = keep;
+          content = keep.join('\n');
+          truncated = true;
+          byteCapped = keep.length < allLines.length;
+        }
+        /* opencode read.ts footer'ları: byte-kesme → offset devam; satır-kesme →
+           offset devam; aksi → dosya sonu. Model BAŞTAN okuma döngüsüne girmez. */
+        let note;
+        if (byteCapped && slice.length) {
+          note = `(Output capped at 50 KB. Showing lines ${offset}-${offset + slice.length - 1}. Use offset=${offset + slice.length} to continue.)`;
+        } else if (truncated && slice.length) {
+          note = `(Showing lines ${offset}-${offset + slice.length - 1} of ${allLines.length}. Use offset=${offset + slice.length} to continue.)`;
+        } else {
+          note = `(End of file - total ${allLines.length} lines)`;
+        }
+        return JSON.stringify({
+          ok: true,
+          path: abs,
+          totalLines: allLines.length,
+          offset,
+          truncated,
+          note,
+          content,
+        });
+      }
+      case 'write_file': {
+        /* opencode write.ts port: üzerine yazmadan önce eski içerik alınır,
+           sonuçta additions/deletions + UI diffView döner */
+        const abs = safeResolve(String(args.path || ''), cwd);
+        const content = String(args.content ?? '');
+        const existed = fs.existsSync(abs);
+        const before = existed && !fs.statSync(abs).isDirectory() ? fs.readFileSync(abs, 'utf8') : '';
+        fs.mkdirSync(path.dirname(abs), { recursive: true });
+        fs.writeFileSync(abs, content, 'utf8');
+        readCacheDrop(abs); /* sonraki read_file taze içerik okusun */
+        const counts = diffLineCounts(before, content);
+        const out = { ok: true, path: abs, bytes: Buffer.byteLength(content), ...counts };
+        if (ctx.wantDiff) {
+          out.diffView = existed
+            ? { path: abs, ...diffRegion(before, content), ...counts }
+            : { path: abs, before: '', after: capDiffText(content), startLine: 1, ...counts };
+        }
+        return JSON.stringify(out);
+      }
+      case 'list_dir': {
+        /* opencode read.ts dizin modu kuralı: localeCompare sıralı, klasörler
+           `/` ile, 500 giriş tavanı + pagination notu */
+        const abs = args.path ? safeResolve(String(args.path), cwd) : cwd;
+        if (!fs.existsSync(abs)) return JSON.stringify({ ok: false, error: `File not found: ${abs}` });
+        if (!fs.statSync(abs).isDirectory()) return JSON.stringify({ ok: false, error: `Path is a file, not a directory: ${abs}` });
+        const all = fs.readdirSync(abs, { withFileTypes: true });
+        const total = all.length;
+        const entries = all.slice(0, 500);
+        const rows = entries.map((e) => {
+          try {
+            const st = fs.statSync(path.join(abs, e.name));
+            return `${e.isDirectory() ? 'd' : '-'} ${String(st.size).padStart(10)} ${e.isDirectory() ? e.name + '/' : e.name}`;
+          } catch {
+            return `-          ? ${e.name}`;
+          }
+        });
+        rows.sort((a, b) => a.slice(13).localeCompare(b.slice(13)));
+        return JSON.stringify({
+          ok: true,
+          path: abs,
+          count: rows.length,
+          ...(total > 500 ? { note: `(Showing 500 of ${total} entries. Use read_file with offset to paginate.)` } : {}),
+          entries: rows.join('\n'),
+        });
+      }
+      case 'grep': {
+        /* opencode grep.ts BİREBİR port: case-sensitive default (rg kuralı),
+           100 match limit, gruplu çıktı `<abs>:` + `  Line N: text`,
+           2000 karakter satır tavanı + truncation notu */
+        const pattern = String(args.pattern || '');
+        let re;
+        try {
+          re = new RegExp(pattern, args.case_insensitive || args.caseInsensitive ? 'i' : '');
+        } catch (e) {
+          return JSON.stringify({ ok: false, error: 'geçersiz regex: ' + String((e && e.message) || e) });
+        }
+        const root = args.path ? safeResolve(String(args.path), cwd) : cwd;
+        if (!fs.existsSync(root)) return JSON.stringify({ ok: false, error: 'yol bulunamadı: ' + root });
+        const incRe = globToRegExp(String(args.include || ''));
+        const hits = []; // { file, line, text }
+        const MAX_MATCHES = 100;
+        walkFiles(root, incRe, (file) => {
+          if (hits.length >= MAX_MATCHES) return true; // dur
+          let text = '';
+          try {
+            const st = fs.statSync(file);
+            if (st.size > 2 * 1024 * 1024) return false; // dev dosyayı atla
+            text = fs.readFileSync(file, 'utf8');
+          } catch {
+            return false;
+          }
+          const lines = text.split('\n');
+          for (let i = 0; i < lines.length && hits.length < MAX_MATCHES; i++) {
+            if (re.test(lines[i])) {
+              const t = lines[i].slice(0, 2000) + (lines[i].length > 2000 ? '...' : '');
+              hits.push({ file, line: i + 1, text: t });
+            }
+          }
+          return false;
+        });
+        /* opencode çıktı formatı: başlık + dosya gruplama + `  Line N: text` */
+        const groups = new Map();
+        for (const h of hits) {
+          if (!groups.has(h.file)) groups.set(h.file, []);
+          groups.get(h.file).push(`  Line ${h.line}: ${h.text}`);
+        }
+        const out = hits.length
+          ? `Found ${hits.length} matches` +
+            (hits.length >= MAX_MATCHES ? ' (more matches available)' : '') +
+            '\n\n' +
+            [...groups.entries()].map(([f, rows]) => `${f}:\n${rows.join('\n')}`).join('\n\n')
+          : 'No files found';
+        return JSON.stringify({
+          ok: true,
+          pattern,
+          count: hits.length,
+          capped: hits.length >= MAX_MATCHES,
+          ...(hits.length >= MAX_MATCHES ? { note: '(Results truncated. Consider using a more specific path or pattern.)' } : {}),
+          matches: out,
+        });
+      }
+      case 'glob': {
+        /* opencode glob.ts BİREBİR port: 100 sonuç limiti + truncation notu,
+           "No files found", path-dosya hatası */
+        const pat = String(args.pattern || '');
+        if (!pat.trim()) return JSON.stringify({ ok: false, error: 'pattern boş olamaz' });
+        const root = args.path ? safeResolve(String(args.path), cwd) : cwd;
+        if (!fs.existsSync(root)) return JSON.stringify({ ok: false, error: 'yol bulunamadı: ' + root });
+        if (fs.statSync(root).isFile()) {
+          return JSON.stringify({ ok: false, error: `glob path must be a directory: ${root}` });
+        }
+        const re = globToRegExp(pat);
+        const LIMIT = 100;
+        const out = [];
+        walkFiles(root, null, (file) => {
+          const rel = path.relative(root, file).split(path.sep).join('/');
+          if (re.test(rel) || re.test(path.basename(file))) {
+            out.push(path.join(root, rel));
+          }
+          return out.length >= LIMIT; // dur
+        });
+        if (!out.length) return JSON.stringify({ ok: true, pattern: pat, count: 0, files: [], note: 'No files found' });
+        return JSON.stringify({
+          ok: true,
+          pattern: pat,
+          count: out.length,
+          files: out,
+          ...(out.length >= LIMIT ? { note: '(Results are truncated: showing first 100 results. Consider using a more specific path or pattern.)' } : {}),
+        });
+      }
+      case 'web_search': {
+        const q = String(args.query || '');
+        const n = Number(args.max_results) || 8;
+        /* sıralı zincir (searxng/tinyfish/python — dahili tarayıcı engine
+           tarafındaki hook'tan gelir); Exa KALDIRILDI */
+        const r = await searchChainWeb(q, n, { signal: ctx.signal });
+        return JSON.stringify(r);
+      }
+      case 'deep_search': {
+        /* hook'suz bağlamda bile çalışsın: yalnız arama zinciri (sayfa okuma yok).
+           Gizli tarayıcı okuması engine.research hook'undan gelir (main process). */
+        const r = await research.deepSearch(
+          args,
+          { search: (q) => searchChainWeb(q, 10, { signal: ctx.signal }) },
+          ctx.signal
+        );
+        return JSON.stringify(r);
+      }
+      case 'python_run': {
+        const t0 = Date.now();
+        /* önce parametreler — interpreter araması gerektirmeyen hatalar erken dönsün */
+        const cliArgs = Array.isArray(args.args)
+          ? args.args.map(String)
+          : args.args
+            ? [String(args.args)]
+            : [];
+        let scriptPath;
+        let madeTemp = false;
+        if (String(args.code || '').trim()) {
+          scriptPath = path.join(
+            os.tmpdir(),
+            `beast-${Date.now().toString(36)}-${Math.random().toString(36).slice(2, 7)}.py`
+          );
+          fs.writeFileSync(scriptPath, String(args.code), 'utf8');
+          madeTemp = true;
+        } else if (String(args.script || '').trim()) {
+          const raw = String(args.script);
+          scriptPath = path.isAbsolute(raw) ? raw : path.join(pythonScriptsDir(), raw);
+        } else {
+          return JSON.stringify({ ok: false, error: 'code ya da script parametresi gerekli' });
+        }
+        if (!fs.existsSync(scriptPath)) {
+          return JSON.stringify({
+            ok: false,
+            error: `script bulunamadı: ${scriptPath}`,
+            hint: `scripts klasörü: ${pythonScriptsDir()}`,
+          });
+        }
+        let probe;
+        try {
+          probe = await ensurePython({ allowDownload: ctx.allowDownload !== false, signal: ctx.signal });
+        } catch (e) {
+          return JSON.stringify({
+            ok: false,
+            error: String((e && e.message) || e),
+            hint: 'BEAST_PYTHON env ile interpreter yolu verilebilir; ya da python.org kur',
+          });
+        }
+        const r = await runPy(probe.exe, scriptPath, cliArgs, cwd, Number(args.timeout_ms) || 120000, ctx.signal);
+        if (madeTemp) { try { fs.unlinkSync(scriptPath); } catch {} }
+        return JSON.stringify({ ok: r.ok, python: probe.source, exitCode: r.code, ms: Date.now() - t0, output: r.output });
+      }
+      case 'http_fetch':
+      case 'webfetch': {
+        /* opencode webfetch.ts port: format (text|markdown|html) + timeout
+           (default 30s, max 120s) — SSRF koruması Beast'te kalır */
+        const format = String(args.format || 'text').toLowerCase();
+        if (!['text', 'markdown', 'html'].includes(format)) {
+          return JSON.stringify({ ok: false, error: 'format: text | markdown | html' });
+        }
+        const timeoutRaw = Number(args.timeout);
+        if (args.timeout != null && (!Number.isFinite(timeoutRaw) || timeoutRaw <= 0)) {
+          return JSON.stringify({ ok: false, error: `Invalid timeout value: ${args.timeout}. Timeout must be a positive number.` });
+        }
+        /* opencode saniye cinsinden bekler; ms gelirse de tolere et */
+        let timeoutMs = 30000;
+        if (Number.isFinite(timeoutRaw) && timeoutRaw > 0) {
+          timeoutMs = Math.min(timeoutRaw <= 120 ? timeoutRaw * 1000 : timeoutRaw, 120000);
+        }
+        const r = await httpFetch(String(args.url || ''), {
+          maxChars: Number(args.max_chars) || MAX_FETCH_CHARS,
+          format,
+          timeoutMs,
+          signal: ctx.signal,
+        });
+        return JSON.stringify(r);
+      }
+      default:
+        return JSON.stringify({ ok: false, error: `unknown tool ${name}` });
+    }
+  } catch (e) {
+    return JSON.stringify({ ok: false, error: String(e && e.message ? e.message : e) });
+  }
+}
+
+/* ---------- Sıralı arama zinciri (Ayarlar → Web Arama'dan değiştirilir) ----------
+   Motorlar: searxng (yerel 8888) · browser (dahili Chromium → Google) ·
+   DuckDuckGo) · tinyfish (API, anahtar gerekir) · python (ddgs/DDG+Bing+Mojeek).
+   Sıra + aç/kapa ayarı settings.json'da (searchChain) saklanır; eski
+   varsayılan AKTİF. Tarayıcı CAPTCHA/trafik verirse 10 dk atlanır. */
+const SEARCH_ENGINE_IDS = ['searxng', 'stealth', 'browser', 'tinyfish', 'python'];
+const DEFAULT_SEARCH_CHAIN = SEARCH_ENGINE_IDS.map((id) => ({ id, on: true }));
+
+let _searchChain = DEFAULT_SEARCH_CHAIN.map((x) => ({ ...x }));
+let _browserBanUntil = 0;
+
+function normalizeSearchChain(list) {
+  const arr = Array.isArray(list) ? list : [];
+  const rows = [];
+  const seen = new Set();
+  for (const item of arr) {
+    const id = String((item && item.id) || item || '').trim();
+    if (!SEARCH_ENGINE_IDS.includes(id) || seen.has(id)) continue;
+    seen.add(id);
+    rows.push({ id, on: !(item && item.on === false) });
+  }
+  for (const id of SEARCH_ENGINE_IDS) {
+    if (!seen.has(id)) rows.push({ id, on: true }); /* listede eksik motor varsayılan AÇIK */
+  }
+  if (!rows.some((r) => r.on)) {
+    const b = rows.find((r) => r.id === 'browser');
+    if (b) b.on = true;
+  }
+  return rows;
+}
+
+function setSearchChain(list) {
+  _searchChain = normalizeSearchChain(list);
+  return _searchChain;
+}
+
+
+function getSearchChain() {
+  return _searchChain.map((x) => ({ ...x }));
+}
+
+function browserBanned() {
+  return Date.now() < _browserBanUntil;
+}
+
+function banBrowser(minutes = 10) {
+  _browserBanUntil = Date.now() + minutes * 60 * 1000;
+}
+
+/* Zinciri sırayla koştur: her motor boş/hata dönerse sıradakine geç.
+   browser motoru yalnız engine'den gelir (hook); araç-bağımsız çağrılarda atlanır. */
+async function searchChainWeb(query, maxResults, { signal, browser } = {}) {
+  const q = String(query || '').trim().slice(0, 400);
+  if (!q) return { ok: false, error: 'boş sorgu' };
+  const cap = Math.max(1, Math.min(12, Number(maxResults) || 8));
+  let out = null;
+  const banned = browserBanned();
+  /* OTOMATİK SEÇİM: yerel SearXNG ayaktaysa (kullanıcı kapatmadıysa) zincirde
+     nerede olursa olsun ÖNE alınır — ücretsiz + CAPTCHAsız + en hızlı aday.
+     isUp() cache'li: kapalıysa ağır maliyeti yok, normal sıra işler. */
+  const rows = _searchChain.slice();
+  try {
+    const i = rows.findIndex((r) => r.id === 'searxng');
+    if (i > 0 && rows[i].on && (await searxng.isUp())) rows.unshift(rows.splice(i, 1)[0]);
+  } catch {}
+  for (const row of rows) {
+    if (out && out.ok && (out.results || []).length) break;
+    if (!row.on) continue;
+    try {
+      if (row.id === 'browser') {
+        if (typeof browser !== 'function' || banned) continue;
+        out = await browser();
+        if (!out || !out.ok || !(out.results || []).length) {
+          banBrowser(); /* tarayıcı sorunlu — 10 dk atla, alternatifler devrede */
+          out = null;
+        }
+      } else if (row.id === 'stealth') {
+        /* obscura yaklaşımı: Chrome TLS taklidi (curl_cffi) */
+        out = await stealthSearch(q, { maxResults: cap, signal });
+      } else if (row.id === 'tinyfish') {
+        out = await tinyfishSearch(q, cap, signal);
+      } else if (row.id === 'searxng') {
+        /* yerel SearXNG — ayakta değilse hızlıca atlanır (probe cache'li) */
+        out = await searxng.search(q, { maxResults: cap, signal });
+      } else if (row.id === 'python') {
+        out = await webSearchFast(q, { maxResults: cap, signal });
+      }
+    } catch {
+      out = null;
+    }
+    if (out && out.ok && !(out.results || []).length) out = null; /* boş → sıradaki motor */
+  }
+  if (out && out.ok && banned && (out.results || []).length) {
+    out.note = 'dahili tarayıcı 10 dk askıda (CAPTCHA/trafik) — alternatif motor kullanıldı';
+  }
+  return out || { ok: false, error: 'web arama başarısız — tüm motorlar boş döndü' };
+}
+
+/* ---------- TinyFish (anahtar girilirse zincirdeki kendi sırasında) ----------
+   GET https://api.search.tinyfish.ai?query=...  ·  Header: X-API-Key
+   Anahtar yoksa bu motor otomatik atlanır; sıradaki motor devreye girer. */
+let _tinyfishKey = null;
+
+function setTinyfishKey(key) {
+  _tinyfishKey = String(key || '').trim() || null;
+}
+
+async function tinyfishSearch(query, maxResults, signal) {
+  if (!_tinyfishKey) return null;
+  try {
+    const sig =
+      signal && typeof AbortSignal !== 'undefined' && AbortSignal.any
+        ? AbortSignal.any([signal, AbortSignal.timeout(20000)])
+        : AbortSignal.timeout(20000);
+    const r = await fetch(
+      `https://api.search.tinyfish.ai?query=${encodeURIComponent(String(query || ''))}`,
+      {
+        headers: {
+          'X-API-Key': _tinyfishKey,
+          'X-TF-Request-Origin': 'api',
+          'X-TF-Client-Name': 'tinyfish-api-key-page',
+        },
+        signal: sig,
+      }
+    );
+    if (!r.ok) return null;
+    const j = await r.json();
+    const results = (j.results || [])
+      .map((x) => ({
+        title: String(x.title || ''),
+        url: String(x.url || ''),
+        snippet: String(x.snippet || '').replace(/\s+/g, ' ').slice(0, 400),
+        engine: 'tinyfish',
+      }))
+      .filter((x) => x.url && x.title);
+    if (!results.length) return null;
+    return { ok: true, engine: 'tinyfish', query, results: results.slice(0, Math.max(1, Math.min(12, Number(maxResults) || 8))) };
+  } catch {
+    return null;
+  }
+}
+
+module.exports = {
+  definitions,
+  exec,
+  /* opencode edit.ts portu (test + dış kullanım) */
+  ocReplace,
+  diffLineCounts,
+  diffRegion,
+  ocReplacers: {
+    SimpleReplacer,
+    LineTrimmedReplacer,
+    BlockAnchorReplacer,
+    WhitespaceNormalizedReplacer,
+    IndentationFlexibleReplacer,
+    EscapeNormalizedReplacer,
+    TrimmedBoundaryReplacer,
+    ContextAwareReplacer,
+    MultiOccurrenceReplacer,
+  },
+  runCommand,
+  runShellCommand,
+  runBashCommand,
+  gitbash,
+  disposeShellSessions,
+  boundToolOutput,
+  globToRegExp,
+  walkFiles,
+  assertPublicHttpUrl,
+  parseDdgResults,
+  htmlToText,
+  webSearch,
+  httpFetch,
+  webSearchFast,
+  seedScript,
+  bundledScriptPath,
+  /* arama zinciri (sıra + aç/kapa Ayarlar → Web Arama'dan) */
+  SEARCH_ENGINE_IDS,
+  DEFAULT_SEARCH_CHAIN,
+  setSearchChain,
+  getSearchChain,
+  searchChainWeb,
+  browserBanned,
+  banBrowser,
+  setTinyfishKey,
+  tinyfishSearch,
+  stealthSearch,
+  /* python altyapısı */
+  ensurePython,
+  findSystemPython,
+  installEmbeddedPython,
+  pythonScriptsDir,
+  embeddedPythonExe,
+  runPy,
+};