@mrgoonie/multix 0.13.0 → 0.14.0

package/CHANGELOG.md

@@ -5,6 +5,18 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.14.0](https://github.com/mrgoonie/multix-cli/compare/v0.13.0...v0.14.0) (2026-10-11)
+
+
+### Added
+
+* **doc:** read convert input from piped stdin ([#84](https://github.com/mrgoonie/multix-cli/issues/84)) ([4596f7f](https://github.com/mrgoonie/multix-cli/commit/4596f7f09da2e8e7a95582296c220f4e007e3171))
+
+
+### Fixed
+
+* **heygen:** harden base URL, output downloads and multi-output paths ([#83](https://github.com/mrgoonie/multix-cli/issues/83)) ([fea7a95](https://github.com/mrgoonie/multix-cli/commit/fea7a950bdf37836a5344301829b028efe4c9a6d))
+
 ## [0.13.0](https://github.com/mrgoonie/multix-cli/compare/v0.12.0...v0.13.0) (2026-10-11)
 
 

package/README.md

@@ -71,7 +71,7 @@ Set at least one provider key. Add to `.env` in your project root or `~/.multix/
 | `FAL_IMAGE_MODEL` | No | Default fal image model (default `fal-ai/flux/schnell`) |
 | `FAL_VIDEO_MODEL` | No | Default fal text-to-video model (default `fal-ai/kling-video/v1.6/standard/text-to-video`) |
 | `HEYGEN_API_KEY` | For HeyGen | [HeyGen API settings](https://app.heygen.com/developers/api). API usage is billed separately from HeyGen web plans |
-| `HEYGEN_BASE_URL` | No | Override the HeyGen API base (default `https://api.heygen.com`; proxies/tests) |
+| `HEYGEN_BASE_URL` | No | Override the HeyGen API base (default `https://api.heygen.com`; proxies/tests). Must be `https://`, except `http://` on `localhost`/`127.0.0.1`/`[::1]` |
 | `FAL_VIDEO_IMAGE_MODEL` | No | Default fal image-to-video model, used when `--image-url` is passed without `-m` (default `fal-ai/kling-video/v1.6/standard/image-to-video`) |
 | `ATLASCLOUD_API_KEY` | For Atlas Cloud | [Atlas Cloud](https://www.atlascloud.ai/console/api-keys) |
 | `ATLASCLOUD_BASE_URL` | No | Override Atlas Cloud media API base (default `https://api.atlascloud.ai/api/v1`) |
@@ -581,7 +581,7 @@ multix heygen video --avatar <look-id> --audio ./voiceover.mp3
 
 # Translation (one job per language; all are downloaded), lipsync, TTS
 multix heygen languages
-multix heygen translate ./talk.mp4 --languages "Spanish, Chinese (Cantonese, Traditional)" [--mode speed|precision] [--output <dir>]
+multix heygen translate ./talk.mp4 --languages "Spanish, Chinese (Cantonese, Traditional)" [--mode speed|precision] [--output <dir|dub.mp4>]   # dub.mp4 -> dub-Spanish.mp4, ...
 multix heygen lipsync --video <ref> --audio <ref>
 multix heygen tts "Welcome to the show." --voice <voice-id> [--output hello.mp3]
 
@@ -624,13 +624,17 @@ multix media batch --input-dir <dir> --output-dir <dir> [--quality 85] [--max-wi
 ### `multix doc`
 
 ```bash
-multix doc convert --input <files...> [--output <path>] [--auto-name] [--model gemini-3.6-flash] [--prompt <str>] [-v]
+multix doc convert --input <files...> [--output <path>] [--auto-name] [--stdin-ext <ext>] [--model gemini-3.6-flash] [--prompt <str>] [-v]
+
+# Read piped stdin with "-" (PDF, images and text auto-detected; else pass --stdin-ext, e.g. pdf; "-" only once, ./- for a file named -)
+cat report.pdf | multix doc convert --input - --output report.md
 ```
 
 Converts PDFs, DOCX/XLSX/PPTX, images, and HTML/text files to Markdown via Gemini.
 Multiple inputs are concatenated with `---` separators into a single output.
+Piped stdin is labelled `<stdin>` in the output and error messages; its temp file is removed when the command ends.
 
-**Default output:** `./multix-output/document-extraction.md`
+**Default output:** `./multix-output/document-extraction.md` (with `--auto-name`: `<input-basename>-extraction.md`, or `stdin-extraction.md` for piped input)
 
 ## Output
 

package/dist/cli.js

@@ -1428,14 +1428,14 @@ function detectThumbUrl(obj) {
   }
   return null;
 }
-async function downloadThumbBeside(thumbUrl, videoPath, logger) {
+async function downloadThumbBeside(thumbUrl, videoPath, logger, download = downloadFile) {
   const match = thumbUrl.match(IMAGE_EXT_RE);
   const ext = match?.[1] ? match[1].toLowerCase().replace("jpeg", "jpg") : "jpg";
   const dir = path6.dirname(videoPath);
   const base = path6.basename(videoPath, path6.extname(videoPath));
   const dest = path6.join(dir, `${base}_thumb.${ext}`);
   try {
-    await downloadFile(thumbUrl, dest);
+    await download(thumbUrl, dest);
     logger?.success(`Saved thumb ${dest}`);
     return dest;
   } catch (e) {
@@ -1450,7 +1450,7 @@ async function maybeDownloadThumb(source, videoPath, opts = {}) {
     opts.logger?.debug("No thumbnail URL detected in provider response");
     return null;
   }
-  const saved = await downloadThumbBeside(url, videoPath, opts.logger);
+  const saved = await downloadThumbBeside(url, videoPath, opts.logger, opts.download);
   if (saved && opts.copyTo) {
     const copyExt = path6.extname(saved);
     const copyBase = path6.basename(opts.copyTo, path6.extname(opts.copyTo));
@@ -4611,11 +4611,15 @@ function trackTempDir(dir) {
   for (const s of SIGNALS) process.once(s, onSignal);
   return dispose;
 }
+var DEFAULT_FLAG = "--files";
+var DEFAULT_EXAMPLE = "cat clip.mp3 | multix gemini transcribe --files -";
+var DEFAULT_EXT_EXAMPLE = "mp3";
 async function stdinToTempFile(opts = {}) {
   const input = opts.input ?? process.stdin;
+  const flag = opts.flag ?? DEFAULT_FLAG;
   if (input.isTTY) {
     throw new ValidationError(
-      "`--files -` reads piped input, but stdin is a terminal. Example: cat clip.mp3 | multix gemini transcribe --files -"
+      `\`${flag} -\` reads piped input, but stdin is a terminal. Example: ${opts.example ?? DEFAULT_EXAMPLE}`
     );
   }
   const forcedExt = opts.ext ? normalizeExt(opts.ext) : void 0;
@@ -4624,7 +4628,8 @@ async function stdinToTempFile(opts = {}) {
   const cleanup = trackTempDir(dir);
   const rawPath = path23.join(dir, "input");
   const hint = setTimeout(() => {
-    process.stderr.write("Waiting for data on stdin (`--files -`)... press Ctrl+C to cancel\n");
+    process.stderr.write(`Waiting for data on stdin (\`${flag} -\`)... press Ctrl+C to cancel
+`);
   }, WAIT_HINT_MS);
   hint.unref?.();
   let head = Buffer.alloc(0);
@@ -4651,7 +4656,7 @@ async function stdinToTempFile(opts = {}) {
       clearTimeout(hint);
       fs20.closeSync(fd);
     }
-    if (size === 0) throw new ValidationError("`--files -` was given but stdin was empty");
+    if (size === 0) throw new ValidationError(`\`${flag} -\` was given but stdin was empty`);
     const detectedExt = forcedExt ?? detectExtension(head);
     const finalPath = path23.join(dir, `stdin${detectedExt ?? ".bin"}`);
     fs20.renameSync(rawPath, finalPath);
@@ -4665,10 +4670,11 @@ function labelError(message, input) {
   return input.path === input.label ? message : message.split(input.path).join(input.label);
 }
 async function resolveFileInputs(files, opts = {}) {
+  const flag = opts.flag ?? DEFAULT_FLAG;
   const stdinCount = files.filter((f) => f === STDIN_ARG).length;
-  if (stdinCount > 1) throw new ValidationError("`-` (stdin) can only be passed once to --files");
+  if (stdinCount > 1) throw new ValidationError(`\`-\` (stdin) can only be passed once to ${flag}`);
   if (opts.stdinExt && stdinCount === 0) {
-    throw new ValidationError("--stdin-ext only applies when --files includes `-`");
+    throw new ValidationError(`--stdin-ext only applies when ${flag} includes \`-\``);
   }
   for (const f of files) {
     if (f !== STDIN_ARG && !fs20.existsSync(f)) throw new ValidationError(`File not found: ${f}`);
@@ -4680,12 +4686,14 @@ async function resolveFileInputs(files, opts = {}) {
   const stdin = await stdinToTempFile({
     input: opts.input,
     ext: opts.stdinExt,
-    maxBytes: opts.maxBytes
+    maxBytes: opts.maxBytes,
+    flag: opts.flag,
+    example: opts.example
   });
   if (!stdin.detected) {
     stdin.cleanup();
     throw new ValidationError(
-      "Could not detect the type of piped stdin; pass --stdin-ext (e.g. --stdin-ext mp3)"
+      `Could not detect the type of piped stdin; pass --stdin-ext (e.g. --stdin-ext ${opts.extExample ?? DEFAULT_EXT_EXAMPLE})`
     );
   }
   return {
@@ -6072,8 +6080,10 @@ var HEYGEN_OPERATIONS = {
 };
 
 // src/providers/heygen/node-helpers.ts
+import { once as once2 } from "events";
 import fs29 from "fs";
 import path32 from "path";
+import { pipeline as pipeline2 } from "stream/promises";
 init_errors();
 init_redact();
 
@@ -6503,6 +6513,7 @@ function videoState(video, kind = "video", id = video.id, opts = {}) {
   return state;
 }
 function translationState(t) {
+  const audioOnly = !str(t.video_url) && !!str(t.audio_url);
   return {
     kind: "translation",
     id: t.id,
@@ -6510,6 +6521,7 @@ function translationState(t) {
     done: t.status === "completed" || t.status === "failed",
     succeeded: t.status === "completed",
     outputUrl: str(t.video_url) ?? str(t.audio_url),
+    ...audioOnly ? { outputIsAudio: true } : {},
     captionUrl: str(t.srt_caption_url) ?? str(t.vtt_caption_url),
     outputLanguage: str(t.output_language),
     failureMessage: str(t.failure_message),
@@ -6632,13 +6644,29 @@ function getJobState(client, kind, id, opts = {}) {
   return createJobStatusReader(client, kind, id, opts)();
 }
 var DEFAULT_POLL_INTERVAL_MS = 1e4;
-function waitForJob(client, kind, id, opts) {
-  const { burnCaptions, ...poll2 } = opts;
+var STAGGER_STEP_MS = 500;
+var STAGGER_MAX_MS = 5e3;
+function staggerOffsetMs(index, intervalMs) {
+  if (!Number.isFinite(index) || index <= 0) return 0;
+  const step = Math.min(STAGGER_STEP_MS, Math.max(0, intervalMs));
+  return Math.min(index * step, STAGGER_MAX_MS);
+}
+var defaultSleep3 = (ms) => new Promise((resolve) => setTimeout(resolve, ms));
+async function waitForJob(client, kind, id, opts) {
+  const { burnCaptions, startDelayMs = 0, ...poll2 } = opts;
+  let pendingDelay = Math.min(Math.max(0, startDelayMs), Math.floor(poll2.timeoutMs / 2));
+  const read = createJobStatusReader(client, kind, id, { burnCaptions });
   return pollUntil({
     intervalMs: DEFAULT_POLL_INTERVAL_MS,
     maxIntervalMs: 3e4,
     ...poll2,
-    fetchStatus: createJobStatusReader(client, kind, id, { burnCaptions }),
+    fetchStatus: async (ctx) => {
+      const delay = pendingDelay;
+      if (delay <= 0) return read(ctx);
+      pendingDelay = 0;
+      await (poll2.sleep ?? defaultSleep3)(delay);
+      return read({ remainingMs: Math.max(1, ctx.remainingMs - delay) });
+    },
     isTerminal: (s) => s.done
   });
 }
@@ -6662,6 +6690,7 @@ function resumeCommand2(kind, id, opts = {}) {
 
 // src/providers/heygen/request-builders.ts
 init_errors();
+init_redact();
 function oneOf(value, allowed, flag) {
   if (value === void 0) return void 0;
   if (!allowed.includes(value)) {
@@ -6683,6 +6712,30 @@ function requireHttps(url, flag) {
   if (!/^https:\/\//i.test(url)) throw new ValidationError(`${flag} must be an https:// URL`);
   return url;
 }
+function isHttpsUrl(url) {
+  try {
+    return new URL(url).protocol === "https:";
+  } catch {
+    return false;
+  }
+}
+var LOOPBACK_HOSTS = /* @__PURE__ */ new Set(["localhost", "127.0.0.1", "[::1]"]);
+function checkBaseUrl(raw, variable) {
+  const value = raw.trim();
+  let url;
+  try {
+    url = new URL(value);
+  } catch {
+    throw new ConfigError(`${variable} must be an absolute https:// URL`);
+  }
+  const loopbackHttp = url.protocol === "http:" && LOOPBACK_HOSTS.has(url.hostname);
+  if (url.protocol !== "https:" && !loopbackHttp) {
+    throw new ConfigError(
+      `${variable} must use https:// (got ${redactUrl(value)}); plain http:// is only allowed for localhost, 127.0.0.1, or [::1]`
+    );
+  }
+  return value;
+}
 function classifyMediaRef(value) {
   const v = value.trim();
   if (!v) throw new ValidationError("Media reference must not be empty");
@@ -6948,9 +7001,11 @@ function requireHeygenKey() {
   return key;
 }
 function createHeygenClientFromEnv() {
+  const apiKey = requireHeygenKey();
+  const override = resolveKey("HEYGEN_BASE_URL")?.trim();
   return new HeygenClient({
-    apiKey: requireHeygenKey(),
-    baseUrl: resolveKey("HEYGEN_BASE_URL") ?? DEFAULT_HEYGEN_BASE_URL
+    apiKey,
+    baseUrl: override ? checkBaseUrl(override, "HEYGEN_BASE_URL") : DEFAULT_HEYGEN_BASE_URL
   });
 }
 async function uploadLocalFile(client, filePath, logger) {
@@ -7012,9 +7067,15 @@ function addWaitOptions(cmd, defaultTimeoutMs = DEFAULT_WAIT_TIMEOUT_MS) {
     "--poll-interval <ms>",
     "Initial poll interval (backs off to 30s)",
     String(DEFAULT_POLL_INTERVAL_MS)
-  ).option("--no-download", "Do not download outputs; print output URLs instead").option("--output <path>", "Output file (or directory when several outputs are produced)").option("--captions", "Also download the caption/subtitle file when one is available").option("--no-thumb", "Skip downloading the video thumbnail").option("--json", "Print a JSON summary instead of human-readable output").option("-v, --verbose", "Verbose logging");
-}
-var OUTPUT_EXTS = /* @__PURE__ */ new Set(["mp4", "webm", "mov", "mp3", "wav", "m4a", "aac", "srt", "vtt"]);
+  ).option("--no-download", "Do not download outputs; print output URLs instead").option(
+    "--output <path>",
+    "Output file or directory. A path ending in / or \\, an existing directory, or a path without an extension is a directory; with several outputs, a media file name gets a per-language suffix (dub.mp4 -> dub-Spanish.mp4)"
+  ).option("--captions", "Also download the caption/subtitle file when one is available").option("--no-thumb", "Skip downloading the video thumbnail").option("--json", "Print a JSON summary instead of human-readable output").option("-v, --verbose", "Verbose logging");
+}
+var VIDEO_EXTS = /* @__PURE__ */ new Set(["mp4", "webm", "mov"]);
+var AUDIO_EXTS = /* @__PURE__ */ new Set(["mp3", "wav", "m4a", "aac"]);
+var MEDIA_EXTS = /* @__PURE__ */ new Set([...VIDEO_EXTS, ...AUDIO_EXTS]);
+var OUTPUT_EXTS = /* @__PURE__ */ new Set([...MEDIA_EXTS, "srt", "vtt"]);
 function inferExtension(url, fallback) {
   try {
     const ext = path32.extname(new URL(url).pathname).slice(1).toLowerCase();
@@ -7026,16 +7087,57 @@ function inferExtension(url, fallback) {
 function safeSlug(value) {
   return value.replace(/[^A-Za-z0-9_-]+/g, "-").replace(/^-+|-+$/g, "").slice(0, 40) || "out";
 }
+var MAX_REDIRECTS = 5;
+var isRedirect = (status) => [301, 302, 303, 307, 308].includes(status);
+async function downloadHttpsOnly(url, dest, timeoutMs) {
+  const signal = AbortSignal.timeout(timeoutMs);
+  let current = url;
+  for (let hop = 0; ; hop++) {
+    if (!isHttpsUrl(current)) {
+      throw new HttpError(0, "refusing to follow a redirect to a non-https URL", current);
+    }
+    let response;
+    try {
+      response = await globalThis.fetch(current, { signal, redirect: "manual" });
+    } catch (cause) {
+      throw new HttpError(0, cause instanceof Error ? cause.message : String(cause), current);
+    }
+    if (response.url && !isHttpsUrl(response.url)) {
+      await response.body?.cancel();
+      throw new HttpError(0, "redirected to a non-https URL", response.url);
+    }
+    const location = response.headers.get("location");
+    if (isRedirect(response.status) && location) {
+      await response.body?.cancel();
+      if (hop >= MAX_REDIRECTS) throw new HttpError(0, "too many redirects", url);
+      current = new URL(location, current).href;
+      continue;
+    }
+    if (!response.ok) throw new HttpError(response.status, "(download failed)", current);
+    if (!response.body) throw new HttpError(0, "Response body is null", current);
+    fs29.mkdirSync(path32.dirname(dest), { recursive: true });
+    const fileStream = fs29.createWriteStream(dest);
+    try {
+      await pipeline2(response.body, fileStream);
+    } catch (err2) {
+      if (!fileStream.closed) await once2(fileStream, "close");
+      throw err2;
+    }
+    return;
+  }
+}
 async function downloadOutput(url, dest, resumeHint) {
   if (!url) {
     throw new ValidationError(`HeyGen reported success but returned no output URL. ${resumeHint}`);
   }
-  if (!/^https?:\/\//i.test(url)) {
-    throw new ValidationError(`HeyGen returned an invalid output URL (${redactUrl(url)})`);
+  if (!isHttpsUrl(url)) {
+    throw new ValidationError(
+      `Refusing to download HeyGen output from a non-https URL (${redactUrl(url)})`
+    );
   }
   const tmp = `${dest}.${process.pid}-${Date.now()}.part`;
   try {
-    await downloadFile(url, tmp, 6e5);
+    await downloadHttpsOnly(url, tmp, 6e5);
     fs29.renameSync(tmp, dest);
   } catch (e) {
     fs29.rmSync(tmp, { force: true });
@@ -7052,15 +7154,55 @@ function defaultOutputPath(state, ext) {
   const lang = state.outputLanguage ? `-${safeSlug(state.outputLanguage)}` : "";
   return path32.join(getOutputDir(), `heygen-${state.kind}${lang}-${safeSlug(state.id)}.${ext}`);
 }
-function outputPathFor(state, ext, output, count) {
-  if (!output) return defaultOutputPath(state, ext);
-  if (count > 1) {
-    const dir = path32.resolve(output);
-    fs29.mkdirSync(dir, { recursive: true });
-    return path32.join(dir, path32.basename(defaultOutputPath(state, ext)));
+function statOrUndefined(p) {
+  try {
+    return fs29.statSync(p);
+  } catch {
+    return void 0;
   }
-  return path32.resolve(output);
 }
+function isDirectoryOutput2(output, target) {
+  if (/[\\/]$/.test(output)) return true;
+  const stat = statOrUndefined(target);
+  if (stat) return stat.isDirectory();
+  return path32.extname(target) === "";
+}
+function withMediaExt(file, ext) {
+  const given = path32.extname(file).slice(1).toLowerCase();
+  const mismatch = VIDEO_EXTS.has(given) && AUDIO_EXTS.has(ext) || AUDIO_EXTS.has(given) && VIDEO_EXTS.has(ext);
+  return mismatch ? `${file.slice(0, -(given.length + 1))}.${ext}` : file;
+}
+function outputPathFor(state, ext, output, count) {
+  if (!output) return defaultOutputPath(state, ext);
+  const target = path32.resolve(output);
+  const givenExt = path32.extname(target);
+  const isTemplate = count > 1 && MEDIA_EXTS.has(givenExt.slice(1).toLowerCase());
+  if (isDirectoryOutput2(output, target) || count > 1 && !isTemplate) {
+    fs29.mkdirSync(target, { recursive: true });
+    return path32.join(target, path32.basename(defaultOutputPath(state, ext)));
+  }
+  if (!isTemplate) return withMediaExt(target, ext);
+  const base = path32.basename(target, givenExt);
+  const suffix = safeSlug(state.outputLanguage ?? state.id);
+  return withMediaExt(path32.join(path32.dirname(target), `${base}-${suffix}${givenExt}`), ext);
+}
+function claimPath(dest, state, used) {
+  const ext = path32.extname(dest);
+  const stem = dest.slice(0, dest.length - ext.length);
+  let candidate = dest;
+  for (let n = 1; used.has(candidate.toLowerCase()); n++) {
+    const id = safeSlug(state.id);
+    candidate = n === 1 ? `${stem}-${id}${ext}` : `${stem}-${id}-${n}${ext}`;
+  }
+  used.add(candidate.toLowerCase());
+  return candidate;
+}
+function captionPathFor(dest, capExt) {
+  const stem = path32.join(path32.dirname(dest), path32.basename(dest, path32.extname(dest)));
+  const capDest = `${stem}.${capExt}`;
+  return capDest.toLowerCase() === dest.toLowerCase() ? `${stem}.captions.${capExt}` : capDest;
+}
+var downloadThumbHttpsOnly = (url, dest) => downloadHttpsOnly(url, dest, 12e4);
 function summarizeState(state) {
   const out = {
     kind: state.kind,
@@ -7078,7 +7220,7 @@ function summarizeState(state) {
   if (state.failureMessage) out.failure_message = state.failureMessage;
   return out;
 }
-async function deliverOne(state, opts, count, logger, files) {
+async function deliverOne(state, opts, count, logger, files, usedPaths) {
   const hint = `Refresh with: ${resumeCommand2(state.kind, state.id, opts)}`;
   if (opts.download === false) {
     if (!state.outputUrl) {
@@ -7091,39 +7233,42 @@ async function deliverOne(state, opts, count, logger, files) {
     };
     return { state, files, urls };
   }
-  const ext = inferExtension(state.outputUrl ?? "", "mp4");
-  const dest = outputPathFor(state, ext, opts.output, count);
+  const ext = inferExtension(state.outputUrl ?? "", state.outputIsAudio ? "mp3" : "mp4");
+  const dest = claimPath(outputPathFor(state, ext, opts.output, count), state, usedPaths);
   await downloadOutput(state.outputUrl, dest, hint);
   logger.success(`Saved ${dest}`);
   files.push(dest);
   if (opts.captions && state.captionUrl) {
     const capExt = inferExtension(state.captionUrl, "srt");
-    const capDest = path32.join(
-      path32.dirname(dest),
-      `${path32.basename(dest, path32.extname(dest))}.${capExt}`
-    );
+    const capDest = claimPath(captionPathFor(dest, capExt), state, usedPaths);
     await downloadOutput(state.captionUrl, capDest, hint);
     logger.success(`Saved ${capDest}`);
     files.push(capDest);
   } else if (opts.captions) {
     logger.warn(`No caption file available for ${state.kind} ${state.id}`);
   }
-  if (state.kind === "video" || state.kind === "agent") {
-    const thumb = await maybeDownloadThumb(state.raw, dest, {
-      skip: opts.thumb === false,
-      logger
-    });
-    if (thumb) files.push(thumb);
+  if ((state.kind === "video" || state.kind === "agent") && opts.thumb !== false) {
+    const thumbUrl = detectThumbUrl(state.raw);
+    if (thumbUrl && !isHttpsUrl(thumbUrl)) {
+      logger.warn(`Skipping thumbnail from a non-https URL (${redactUrl(thumbUrl)})`);
+    } else {
+      const thumb = await maybeDownloadThumb(state.raw, dest, {
+        logger,
+        download: downloadThumbHttpsOnly
+      });
+      if (thumb) files.push(thumb);
+    }
   }
   return { state, files };
 }
 async function deliverOutputs(states, opts, logger) {
   const finished = [];
   const failures = [];
+  const usedPaths = /* @__PURE__ */ new Set();
   for (const state of states) {
     const files = [];
     try {
-      finished.push(await deliverOne(state, opts, states.length, logger, files));
+      finished.push(await deliverOne(state, opts, states.length, logger, files, usedPaths));
     } catch (e) {
       failures.push(`${state.kind} ${state.id}: ${e instanceof Error ? e.message : String(e)}`);
       if (files.length > 0) finished.push({ state, files });
@@ -7176,9 +7321,10 @@ async function completeJobs(client, kind, ids, opts, logger) {
   );
   const settled = await Promise.allSettled(
     ids.map(
-      (id) => waitForJob(client, kind, id, {
+      (id, index) => waitForJob(client, kind, id, {
         timeoutMs,
         intervalMs,
+        startDelayMs: staggerOffsetMs(index, intervalMs),
         maxIntervalMs: Math.max(intervalMs, 3e4),
         burnCaptions: opts.burnCaptions,
         resumeHint: `Resume with: ${resumeCommand2(kind, id, opts)}`,
@@ -10079,8 +10225,8 @@ async function downloadOpenRouterVideo(opts) {
   }
   fs43.mkdirSync(path50.dirname(dest), { recursive: true });
   const ws = fs43.createWriteStream(dest);
-  const { pipeline: pipeline3 } = await import("stream/promises");
-  await pipeline3(r.body, ws);
+  const { pipeline: pipeline4 } = await import("stream/promises");
+  await pipeline4(r.body, ws);
   logger.success(`Saved ${dest}`);
   if (opts.output) {
     fs43.mkdirSync(path50.dirname(path50.resolve(opts.output)), { recursive: true });
@@ -10408,7 +10554,7 @@ function assertSucceeded(prediction2) {
     PROVIDER
   );
 }
-var defaultSleep3 = (ms) => new Promise((resolve) => setTimeout(resolve, ms));
+var defaultSleep4 = (ms) => new Promise((resolve) => setTimeout(resolve, ms));
 var ReplicateClient = class {
   apiKey;
   baseUrl;
@@ -10419,7 +10565,7 @@ var ReplicateClient = class {
     this.apiKey = opts.apiKey;
     this.baseUrl = (opts.baseUrl || DEFAULT_REPLICATE_BASE_URL).replace(/\/+$/, "");
     this.doFetch = opts.fetch ?? ((input, init) => globalThis.fetch(input, init));
-    this.sleep = opts.sleep ?? defaultSleep3;
+    this.sleep = opts.sleep ?? defaultSleep4;
   }
   headers(extra = {}) {
     return { accept: "application/json", authorization: `Bearer ${this.apiKey}`, ...extra };
@@ -10610,11 +10756,11 @@ init_errors();
 // src/providers/replicate/node-outputs.ts
 init_errors();
 init_redact();
-import { once as once2 } from "events";
+import { once as once3 } from "events";
 import fs44 from "fs";
 import path51 from "path";
 import { Readable } from "stream";
-import { pipeline as pipeline2 } from "stream/promises";
+import { pipeline as pipeline3 } from "stream/promises";
 
 // src/providers/replicate/output.ts
 init_redact();
@@ -10785,12 +10931,12 @@ async function writeResponse(response, dest) {
   if (!response.body) throw new HttpError(0, "Response body is null", response.url);
   const fileStream = fs44.createWriteStream(dest);
   try {
-    await pipeline2(
+    await pipeline3(
       Readable.fromWeb(response.body),
       fileStream
     );
   } catch (err2) {
-    if (!fileStream.closed) await once2(fileStream, "close");
+    if (!fileStream.closed) await once3(fileStream, "close");
     throw err2;
   }
 }
@@ -11460,7 +11606,6 @@ function registerCheckCommand(program2) {
 }
 
 // src/commands/doc/convert.ts
-init_errors();
 import fs47 from "fs";
 import path53 from "path";
 
@@ -11478,27 +11623,45 @@ Requirements:
 Output only the markdown content without any preamble or explanation.`;
 
 // src/commands/doc/convert.ts
+function autoNameBase(input) {
+  if (input === STDIN_ARG) return "stdin";
+  return path53.basename(input, path53.extname(input));
+}
 async function convertDocuments(opts) {
-  const { inputs, output, autoName = false, verbose, logger } = opts;
+  const { output, autoName = false, verbose, logger } = opts;
   const model = opts.model ?? DOC_MODEL_DEFAULT;
   const prompt = opts.prompt ?? DEFAULT_CONVERT_PROMPT;
-  for (const f of inputs) {
-    if (!fs47.existsSync(f)) throw new ValidationError(`Input file not found: ${f}`);
-  }
-  let outputPath;
-  if (output) {
-    outputPath = path53.resolve(output);
-  } else if (autoName && inputs.length === 1 && inputs[0]) {
-    const base = path53.basename(inputs[0], path53.extname(inputs[0]));
-    outputPath = path53.join(getOutputDir(), `${base}-extraction.md`);
-  } else {
-    outputPath = path53.join(getOutputDir(), "document-extraction.md");
+  const { inputs, cleanup } = await resolveFileInputs(opts.inputs, {
+    stdinExt: opts.stdinExt,
+    input: opts.stdin,
+    flag: "--input",
+    example: "cat report.pdf | multix doc convert --input -",
+    extExample: "pdf"
+  });
+  let failed;
+  try {
+    let outputPath;
+    if (output) {
+      outputPath = path53.resolve(output);
+    } else if (autoName && opts.inputs.length === 1 && opts.inputs[0]) {
+      outputPath = path53.join(getOutputDir(), `${autoNameBase(opts.inputs[0])}-extraction.md`);
+    } else {
+      outputPath = path53.join(getOutputDir(), "document-extraction.md");
+    }
+    fs47.mkdirSync(path53.dirname(outputPath), { recursive: true });
+    failed = await convertResolved(inputs, { outputPath, model, prompt, verbose, logger });
+  } finally {
+    cleanup();
   }
-  fs47.mkdirSync(path53.dirname(outputPath), { recursive: true });
+  if (failed > 0) process.exit(1);
+}
+async function convertResolved(inputs, ctx) {
+  const { outputPath, model, prompt, verbose, logger } = ctx;
   const results = [];
   for (let i = 0; i < inputs.length; i++) {
-    const filePath = inputs[i];
-    logger?.debug(`[${i + 1}/${inputs.length}] Converting: ${filePath}`);
+    const input = inputs[i];
+    const filePath = input.path;
+    logger?.debug(`[${i + 1}/${inputs.length}] Converting: ${input.label}`);
     let markdown;
     let errorMsg;
     for (let attempt = 0; attempt < 3; attempt++) {
@@ -11530,7 +11693,7 @@ async function convertDocuments(opts) {
         markdown = extractText(resp);
         break;
       } catch (e) {
-        errorMsg = e instanceof Error ? e.message : String(e);
+        errorMsg = labelError(e instanceof Error ? e.message : String(e), input);
         if (isGeminiRateLimitError(e)) break;
         if (attempt < 2) {
           const wait = 2 ** attempt * 1e3;
@@ -11539,8 +11702,11 @@ async function convertDocuments(opts) {
         }
       }
     }
+    if (markdown === void 0) {
+      logger?.error(`Failed: ${input.label} \u2014 ${errorMsg}`);
+    }
     results.push({
-      file: filePath,
+      file: input.label,
       status: markdown !== void 0 ? "success" : "error",
       markdown,
       error: errorMsg
@@ -11554,7 +11720,7 @@ async function convertDocuments(opts) {
 `);
   for (const r of results) {
     lines.push(`
-## ${path53.basename(r.file)}
+## ${path53.basename(r.file).replace(/[<>]/g, "\\$&")}
 `);
     if (r.status === "success" && r.markdown) {
       lines.push(`${r.markdown}
@@ -11580,8 +11746,7 @@ ${"=".repeat(50)}`);
       console.log(`  [${r.status}] ${r.file}`);
     }
   }
-  const failed = results.filter((r) => r.status === "error").length;
-  if (failed > 0) process.exit(1);
+  return results.length - succeeded;
 }
 function sleep4(ms) {
   return new Promise((r) => setTimeout(r, ms));
@@ -11594,14 +11759,21 @@ function registerDocCommands(program2) {
   );
   doc.command("convert").description("Convert documents to Markdown using Gemini Files API").requiredOption(
     "--input <files...>",
-    "Input file(s) to convert",
+    "Input file(s) to convert (use - to read piped stdin)",
     (val, prev) => [...prev, val],
     []
-  ).alias("-i").option("--output <path>", "Output markdown file path").option("--auto-name", "Derive output filename from input basename (single file only)").option("--model <id>", `Gemini model to use (default: ${DOC_MODEL_DEFAULT})`).option("--prompt <text>", "Custom conversion prompt").option("-v, --verbose", "Verbose logging").action(
+  ).alias("-i").option(
+    "--stdin-ext <ext>",
+    "Extension for piped stdin when auto-detection fails (e.g. pdf, txt)"
+  ).option("--output <path>", "Output markdown file path").option(
+    "--auto-name",
+    "Derive output filename from input basename (single file only; stdin becomes stdin-extraction.md)"
+  ).option("--model <id>", `Gemini model to use (default: ${DOC_MODEL_DEFAULT})`).option("--prompt <text>", "Custom conversion prompt").option("-v, --verbose", "Verbose logging").action(
     async (opts) => {
       const logger = createLogger({ verbose: opts.verbose ?? false });
       await convertDocuments({
         inputs: opts.input,
+        stdinExt: opts.stdinExt,
         output: opts.output,
         autoName: opts.autoName,
         model: opts.model,
@@ -11623,13 +11795,13 @@ import fs48 from "fs";
 import path54 from "path";
 
 // src/commands/media/detect-type.ts
-var VIDEO_EXTS = /* @__PURE__ */ new Set([".mp4", ".mov", ".avi", ".mkv", ".webm", ".flv", ".mpg", ".mpeg"]);
-var AUDIO_EXTS = /* @__PURE__ */ new Set([".mp3", ".wav", ".m4a", ".flac", ".aac", ".ogg", ".aiff"]);
+var VIDEO_EXTS2 = /* @__PURE__ */ new Set([".mp4", ".mov", ".avi", ".mkv", ".webm", ".flv", ".mpg", ".mpeg"]);
+var AUDIO_EXTS2 = /* @__PURE__ */ new Set([".mp3", ".wav", ".m4a", ".flac", ".aac", ".ogg", ".aiff"]);
 var IMAGE_EXTS2 = /* @__PURE__ */ new Set([".jpg", ".jpeg", ".png", ".webp", ".heic", ".heif", ".gif", ".bmp"]);
 function extToKind(ext) {
   const lower = ext.toLowerCase();
-  if (VIDEO_EXTS.has(lower)) return "video";
-  if (AUDIO_EXTS.has(lower)) return "audio";
+  if (VIDEO_EXTS2.has(lower)) return "video";
+  if (AUDIO_EXTS2.has(lower)) return "audio";
   if (IMAGE_EXTS2.has(lower)) return "image";
   return void 0;
 }
@@ -11805,7 +11977,7 @@ async function optimizeVideo(opts) {
 }
 
 // src/commands/media/batch.ts
-var ALL_EXTS = /* @__PURE__ */ new Set([...VIDEO_EXTS, ...AUDIO_EXTS, ...IMAGE_EXTS2]);
+var ALL_EXTS = /* @__PURE__ */ new Set([...VIDEO_EXTS2, ...AUDIO_EXTS2, ...IMAGE_EXTS2]);
 async function batchOptimize(opts) {
   const {
     inputDir,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mrgoonie/multix",
-  "version": "0.13.0",
+  "version": "0.14.0",
   "description": "AI multimodal CLI — OpenAI, Gemini, MiniMax, OpenRouter, Leonardo, BytePlus, ElevenLabs, fal.ai, Atlas Cloud, Replicate, HeyGen, ffmpeg, ImageMagick, doc-to-md",
   "type": "module",
   "bin": {

package/skill/SKILL.md

@@ -1,7 +1,7 @@
 ---
 name: multix
 description: AI multimodal CLI — generate and edit images (OpenAI, Codex driver, Gemini Nano Banana, Imagen, MiniMax, OpenRouter, Leonardo, BytePlus Seedream, Cloudflare Workers AI FLUX.1 Schnell, fal.ai FLUX, Replicate FLUX), generate video (Gemini Veo, MiniMax Hailuo, OpenRouter via Veo/Seedance/Kling, Leonardo via Motion/Veo/Kling/Hailuo/Seedance/LTX, BytePlus Seedance, Cloudflare AI Gateway via Replicate, fal.ai queue models, Replicate direct API models incl. P-Video, HeyGen Video Agent and scripted avatar videos), video translation and lipsync (HeyGen), TTS/STT/music (OpenAI, Gemini Flash TTS incl. Gemini 3.8 with delivery-style control, MiniMax, Cloudflare Workers AI MeloTTS, ElevenLabs), 3D models (BytePlus Hyper3D/Hitem3d), analyze/transcribe media, convert documents to Markdown, optimize media via ffmpeg/ImageMagick, and self-update the CLI. Use whenever the user wants to create, edit, or transform images/video/audio/3D/documents from a CLI, mentions any of: OpenAI, Codex image driver, image-to-image, i2i, image edit, watercolor/cyberpunk style transfer, reference image, OpenRouter, Nano Banana, Flux, Seedream, Veo, Hailuo, Seedance, Kling, Cloudflare Workers AI, AI Gateway, fal.ai, Atlas Cloud, Replicate, replicate.com models, Replicate predictions, HeyGen, avatar video, talking head, Video Agent, video translation, dubbing, lipsync, video generation, text-to-video, image-to-video, i2v, ElevenLabs voice cloning, updating/upgrading multix, or asks for batch media optimization.
-version: 0.13.0 # x-release-please-version
+version: 0.14.0 # x-release-please-version
 ---
 
 # multix — AI Multimodal CLI
@@ -194,7 +194,7 @@ multix heygen video-agent "45-second explainer for a note app" [--avatar <look-i
 multix heygen avatars [--type photo_avatar] [--cursor <token>]
 multix heygen voices [--language English]
 multix heygen video "Hello!" --avatar <look-id> [--voice <id>] [--aspect-ratio 9:16] [--engine avatar_iv|avatar_v|avatar_iii] [--burn-captions]
-multix heygen translate <ref> --languages "Spanish, Chinese (Cantonese, Traditional)" [--output <dir>]
+multix heygen translate <ref> --languages "Spanish, Chinese (Cantonese, Traditional)" [--output <dir|dub.mp4>]   # several languages: dub.mp4 -> dub-Spanish.mp4, ...
 multix heygen lipsync --video <ref> --audio <ref>
 multix heygen tts "Hello" --voice <voice-id>
 multix heygen template <template-id> --input @vars.json
@@ -354,6 +354,7 @@ multix byteplus 3d --input-image ./front.png ./side.png --prompt "Generate from
 
 # Docs → Markdown (Gemini)
 multix doc convert --input report.pdf [--auto-name] [--output ./out.md]
+cat report.pdf | multix doc convert --input - --output ./out.md   # "-" reads stdin; --stdin-ext <ext> (e.g. pdf) if undetected
 
 # Media optimization (ffmpeg + ImageMagick)
 multix media optimize --input <file> --output <file> [--target-size <MB>] [--quality 85] [--max-width 1920]