@remixmate/cli 0.9.68 → 0.9.70

package/dist/manifest.json

@@ -1,7 +1,7 @@
 {
   "schemaVersion": 1,
-  "version": "0.9.68",
-  "generatedAt": "2026-10-05T12:22:45.024Z",
+  "version": "0.9.70",
+  "generatedAt": "2026-10-06T15:32:49.285Z",
   "skills": [
     {
       "id": "doc-parse",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@remixmate/cli",
-  "version": "0.9.68",
+  "version": "0.9.70",
   "description": "AI media generation skills for Claude Code / Codex — 12 skills covering image, video, voice, digital human, web screenshot, web recording, script, template registry, rendering, Jianying export, and video deconstruction.",
   "type": "module",
   "main": "dist/index.js",

package/skills/render-video/SKILL.md

@@ -117,14 +117,28 @@ python3 <SkillDir>/scripts/render_video.py --job-id 42 --upload-title "My video"
 After rendering, the script auto-uploads by default; `render-manifest.json` then contains an
 `upload.fileUrl` field, and `upload.playbackUrl` (transcoded) once VOD processing finishes.
 
-**Do not retype the video URL into your reply.** On success the script prints a structured
-`__render_video_asset__` line; the host reads the authoritative URL from there and renders a
-player for the user. Tell the user the video is ready — the player appears on its own.
-
-Reproducing a 32-char opaque URL from memory is unreliable: on 2026-08-30 a single character
-was dropped (`…c1c20102` → `…c1c2012`), so the user got a 404 while the file sat fine on the
-CDN. The URL in your prose is redundant with the structured asset and is the only copy that
-can be wrong — the UI now renders it as plain, non-clickable text for exactly this reason.
+On success the script prints these lines on stdout:
+
+```
+🎬 Video playback URL: <videoUrl>
+🖼️ Cover image URL: <coverUrl>          # only when the template ships a cover
+__render_video_asset__ {"url": "...", "coverUrl": "...", "durationSec": ...}
+```
+
+How to report the result depends on the host:
+
+- **Host with a built-in player** (the RemixMate web app; the env has `REMIXMATE_HOST=workbuddy`):
+  do **not** retype the URLs. The host reads `__render_video_asset__` and renders the player and
+  cover itself. Just tell the user the video is ready.
+- **Any other host** (Codex, Claude Code, other CLIs): nothing renders a player, so the links in
+  your reply are the only way the user can get to the result. You **must** output both:
+  - the video link (from `🎬 Video playback URL`)
+  - the cover link (from `🖼️ Cover image URL`, or `coverUrl` in `__render_video_asset__`).
+    If neither is printed, say the template has no cover rather than leaving it out silently.
+
+  Copy each URL **verbatim from the stdout line**, character for character. Never retype one
+  from memory: on 2026-08-30 a single character was dropped (`…c1c20102` → `…c1c2012`), and the
+  user got a 404 while the file sat fine on the CDN.
 
 ### Render but skip the upload
 

package/skills/render-video/scripts/render_video.py

@@ -3059,6 +3059,10 @@ Examples:
             effective_playback = playback_url or remote_url
             if effective_playback and not effective_playback.startswith("vod://"):
                 print(f"\n🎬 Video playback URL: {effective_playback}")
+                # 封面也走 stdout：Codex / Claude Code 等没有内联播放器的宿主只能从
+                # 这里拿到封面地址（之前只打到 stderr，模型回复里因此只有视频没有封面）。
+                if upload_info.get("coverUrl"):
+                    print(f"🖼️ Cover image URL: {upload_info['coverUrl']}")
 
                 # 结构化资产标记 —— 宿主（ab-agent）据此把权威 URL 作为 attachment
                 # 下发前端渲染播放器，绕开模型正文。
@@ -3113,6 +3117,9 @@ Examples:
                     # ab-api 保存 manifest 时按 fileId 给成片建语义卡 / 片段 / 血缘
                     "fileId": upload_result.get("fileId"),
                 }
+                # 本地渲染 + 上传成功时同样把地址放到 stdout，供无播放器的宿主回显给用户
+                if upload_result.get("fileUrl"):
+                    print(f"\n🎬 Video playback URL: {upload_result['fileUrl']}")
                 if manifest_path:
                     save_json(manifest, manifest_path)
                     LogPrint(f"   Manifest updated: {manifest_path}", file=sys.stderr)

package/skills/render-video/version.json

@@ -2,6 +2,6 @@
   "skillName": "render-video",
   "repoName": "agent-skill-media-maker",
   "skillId": "473",
-  "version": "V35",
+  "version": "V36",
   "skillDescription": "Final-render skill: loads a persisted RenderPlan by `job_id` and drives the Remotion engine to produce the final video.\n\nUse this skill as soon as the user mentions any of these intents (after assets are already prepared):\n- Render the video, composite the video, export the video\n- Turn the prepared assets into the final clip\n- Render with Remotion\n\nPrerequisite: assets must already be generated via `prepare_video_assets`. This skill never resolves or regenerates assets — pass it a `job_id` from a previous `prepare_video_assets` call.\n\n⚠️ Stop-and-confirm gate: never call this skill until the user has explicitly confirmed the assets prepared by `prepare_video_assets`. If those assets were prepared in the current turn and the user has not replied since, stop and ask instead of rendering."
 }

package/skills/web-read/SKILL.md

@@ -110,6 +110,8 @@ python3 <ReadScript> --url "https://example.com/x" --user-agent "Mozilla/5.0 (Ma
 
 The stderr diagnostics tell you which case you are in: `⚠️ 整页有 N 字符，却只抽到 M` means the wrong container was chosen (→ `--selector`), while `⚠️ 这个页面几乎没有文本` means the page itself never rendered text (→ waiting, cookies, or user-agent).
 
+Slow or unreachable sub-resources do not fail the read: images / fonts / media / stylesheets are never fetched, every other sub-resource is cut off after 8s, and extraction runs on whatever DOM has loaded (stderr says `[web_read] N 个子资源…已跳过`). A non-zero `打开页面失败` therefore means the server never returned the page itself — retrying the same URL with a bigger timeout, another user-agent, device or browser will not help; ask for the text or a different URL.
+
 ## Links and images
 
 Both are dropped by default, because they are noise for a summarize/rewrite task and they inflate the character budget.

package/skills/web-screenshot/scripts/_media_screenshot/bootstrap.py

@@ -21,11 +21,15 @@ def _ensure_playwright_pkg() -> None:
     _pip_install("playwright")
 
 
-def ensure_chromium() -> None:
-    """Install the chromium browser via `playwright install`."""
-    print("[bootstrap] playwright install chromium ...", file=sys.stderr)
+def ensure_browser(name: str = "chromium") -> None:
+    """Install the given browser engine via `playwright install <name>`.
+
+    Must install the engine that is about to be launched: installing chromium
+    and then relaunching firefox fails the same way a second time.
+    """
+    print(f"[bootstrap] playwright install {name} ...", file=sys.stderr)
     subprocess.check_call(
-        [sys.executable, "-m", "playwright", "install", "chromium"]
+        [sys.executable, "-m", "playwright", "install", name]
     )
 
 

package/skills/web-screenshot/scripts/_media_screenshot/browser.py

@@ -4,7 +4,7 @@ from __future__ import annotations
 from typing import Any
 
 from . import js_loader
-from .bootstrap import ensure_chromium
+from .bootstrap import ensure_browser
 
 
 def pick_browser_type(p, name: str | None):
@@ -42,15 +42,35 @@ def build_context_options(p, cfg: dict, record_video_dir: str | None) -> dict:
     return opts
 
 
+def _is_missing_executable(e: Exception) -> bool:
+    msg = str(e)
+    return "Executable doesn't exist" in msg or "playwright install" in msg.lower()
+
+
 def launch_with_browser_install(p, browser_name: str | None, launch_kwargs: dict | None = None):
     btype = pick_browser_type(p, browser_name)
+    engine = btype.name  # "chromium" | "firefox" | "webkit"
+    try:
+        return btype.launch(**(launch_kwargs or {}))
+    except Exception as e:
+        if not _is_missing_executable(e):
+            raise
+    # 引擎没装：装「这次要启动的那个」引擎再试一次。装不上 / 装了仍起不来时给一句
+    # 能据以行动的话，而不是 30 行 Playwright traceback。
+    hint = (
+        f"浏览器引擎 {engine} 在当前环境不可用。"
+        + ("去掉 browser 参数改用默认的 chromium 即可。" if engine != "chromium"
+           else "请运行 `python3 -m playwright install chromium`。")
+    )
+    try:
+        ensure_browser(engine)
+    except Exception as e:
+        raise SystemExit(f"{hint}\n（自动安装失败：{type(e).__name__}: {str(e).splitlines()[0] if str(e) else ''}）")
     try:
         return btype.launch(**(launch_kwargs or {}))
     except Exception as e:
-        msg = str(e)
-        if "Executable doesn't exist" in msg or "playwright install" in msg.lower():
-            ensure_chromium()
-            return btype.launch(**(launch_kwargs or {}))
+        if _is_missing_executable(e):
+            raise SystemExit(hint)
         raise
 
 

package/skills/web-screenshot/scripts/_media_screenshot/reader.py

@@ -12,12 +12,74 @@ disk so nothing is actually lost.
 from __future__ import annotations
 
 import json
+import sys
 from pathlib import Path
+from urllib.parse import urlsplit
 
 from . import js_loader
 from .bootstrap import ensure_runtime
 from .browser import apply_pre_action_waits, build_context_options, launch_with_browser_install
 
+# 读正文只需要 HTML 和（SPA 场景下）脚本 / 接口；图片、字体、媒体、样式表一律不拉。
+# 抽取逻辑只读 DOM 文本与属性（img 的 src 照样在），不依赖布局。
+_SKIPPED_RESOURCE_TYPES = {"image", "media", "font", "stylesheet"}
+# 单个子资源的拉取上限。浏览器要等 defer 脚本 / 阻塞样式表下载完才触发
+# DOMContentLoaded，一个从本机出口挂死的 CDN 就能让整页永远「打不开」
+# （2026-10-06：新加坡出口读腾讯云社区文章，_next chunk 与广州 COS 资源全部挂起）。
+# 超过上限的请求被 abort，浏览器跳过它继续解析。
+_SUBRESOURCE_TIMEOUT_MS = 8000
+# 拿到主文档响应头之后，最多再等这么久的 DOMContentLoaded；到点不报错，按当前 DOM 抽取。
+_DOM_READY_TIMEOUT_MS = 15000
+
+
+class _SubresourceGate:
+    """page.route 处理器：丢弃不需要的资源，给其余子资源加单请求超时。"""
+
+    def __init__(self) -> None:
+        self.stalled: list[str] = []
+
+    def __call__(self, route) -> None:
+        req = route.request
+        if req.is_navigation_request():
+            route.continue_()
+            return
+        if req.resource_type in _SKIPPED_RESOURCE_TYPES:
+            route.abort()
+            return
+        try:
+            resp = route.fetch(timeout=_SUBRESOURCE_TIMEOUT_MS)
+        except Exception:
+            self.stalled.append(urlsplit(req.url).hostname or "?")
+            try:
+                route.abort()
+            except Exception:
+                pass  # 页面已关闭 / 已导航走
+            return
+        try:
+            route.fulfill(response=resp)
+        except Exception:
+            pass
+
+    def summary(self) -> str:
+        hosts = sorted(set(self.stalled))
+        shown = ", ".join(hosts[:5]) + (" …" if len(hosts) > 5 else "")
+        return f"{len(self.stalled)} 个子资源超过 {_SUBRESOURCE_TIMEOUT_MS // 1000}s 未返回已跳过（{shown}）"
+
+
+def _navigation_error(url: str, e: Exception) -> SystemExit:
+    # 一条模型能据以行动的错误，胜过 30 行 Playwright traceback。
+    # 走到这里说明连主文档的响应头都没拿到（子资源挂死已由 _SubresourceGate 兜住），
+    # 问题在网络层或对方服务器；换参数重试几乎不可能有不同结果，要明确告诉调用方别再试。
+    first = str(e).splitlines()[0] if str(e) else ""
+    kind = "超时未响应" if type(e).__name__ == "TimeoutError" else "连接失败"
+    return SystemExit(
+        f"打开页面失败（{kind}）：{url}\n"
+        f"{type(e).__name__}: {first}\n"
+        "服务器没有返回页面：多半是从当前网络出口无法访问该站点，或站点拒绝了本机 IP。"
+        "加大 timeout、换 user_agent / device / browser 通常不会有不同结果，不要对同一 URL 反复重试；"
+        "请让用户直接粘贴正文，或换一个可访问的链接。"
+    )
+
 
 def extract(cfg: dict) -> dict:
     """Load cfg['url'] and return the raw extraction dict from the page."""
@@ -38,19 +100,24 @@ def extract(cfg: dict) -> dict:
             if cfg.get("timeout"):
                 context.set_default_timeout(cfg["timeout"])
             page = context.new_page()
+            gate = _SubresourceGate()
+            page.route("**/*", gate)
             try:
                 try:
-                    response = page.goto(cfg["url"], wait_until="domcontentloaded")
+                    # commit = 拿到主文档响应头即返回，不让任何子资源决定「页面打不打得开」。
+                    response = page.goto(cfg["url"], wait_until="commit")
                 except Exception as e:
-                    # 一条模型能据以行动的错误，胜过 30 行 Playwright traceback。
-                    # 导航失败是这个 skill 最常见的失败形态（打不开 / 超时 / DNS），
-                    # 让它读起来像"这个网址打不开"，而不是像程序崩了。
-                    raise SystemExit(
-                        f"打开页面失败：{cfg['url']}\n"
-                        f"{type(e).__name__}: {str(e).splitlines()[0]}\n"
-                        "可能是网络不可达、站点屏蔽了无头浏览器，或加载超过了超时时间。"
-                        "可尝试加大 --timeout、换 --user-agent，或确认该 URL 在本机能打开。"
-                    )
+                    raise _navigation_error(cfg["url"], e)
+                dom_ready = True
+                try:
+                    page.wait_for_load_state("domcontentloaded", timeout=_DOM_READY_TIMEOUT_MS)
+                except Exception:
+                    dom_ready = False
+                if gate.stalled or not dom_ready:
+                    note = gate.summary() if gate.stalled else "部分子资源未返回"
+                    if not dom_ready:
+                        note += f"；页面 {_DOM_READY_TIMEOUT_MS // 1000}s 内未完成解析，按当前已加载的 DOM 抽取"
+                    print(f"[web_read] {note}", file=sys.stderr)
                 status = response.status if response else None
                 if status is not None and status >= 400:
                     # Keep going: many sites serve real content under a 403/404

package/skills/web-screenshot/version.json

@@ -2,6 +2,6 @@
   "skillName": "web-screenshot",
   "repoName": "agent-skill-media-maker",
   "skillId": "555",
-  "version": "V7",
+  "version": "V8",
   "skillDescription": "Web-page screenshot skill (still images only, png/jpg), built on Playwright Python — captures any URL to a local image file.\nSupports: full-page / viewport / element / region capture, device emulation, waiting for dynamic content, hiding or masking distracting elements, and static annotations (highlight box / arrow / caption / numbered sequence / redact / code-line highlight).\n\nImages only (`.png` / `.jpg`). For video (`.webm` / screen recording / storyboard clips) use web-record / `web_record`.\n\nUse this skill immediately whenever the user asks for any of:\n- Web page screenshot, screen capture, full-page screenshot, long screenshot\n- Capture a specific element / region, partial screenshot, focus on an area\n- Hide or mask elements before capturing (cookie banner, sensitive information)\n- Annotate a still screenshot (highlight box / arrow / label / caption bar / numbering)\n- Highlight lines L5-L20 of a file and screenshot it\n- Wait for an element to appear, then capture\n- Mobile / device-emulated screenshot (iPhone / Pixel, etc.)\n- Capture with cookies / a logged-in session / storageState\n\nEven when the user does not say \"playwright\", any request to screenshot a URL and get a local image file should route here."
 }