Video Podcast Maker
Topic to 4K video podcast: research, script, TTS, Remotion render, BGM mix
Test report
- Verdict
- Works with setup
- Score
- Tested
- Jul 21, 2026
- Environment
- Claude Code 2.x (agent harness)
- Upstream re-checked
- Aug 10, 2026 · 1502fb5
I downloaded the skill's real scripts and ran them. `check_prereqs.py` printed `ALL_OK (backend=edge)` and `migrate_prefs.py --dry-run` printed `[dry-run] Would create user_prefs.json at v1.6 from template`, so the Python side genuinely works. For output I wrote a baseline "why HTTP/3 uses QUIC" script before reading the body (4 scenes, 60s, ad-hoc markdown, no machine-readable format), then followed references/workflow-script.md Steps 1-4 and got topic_definition.md plus a podcast.txt with 8 `[SECTION:xxx]` blocks, density tiers mapped to the named component library (ComparisonCard/FlowChart/DataBar), and the platform+language outro from the skill's CTA table; the skill's own `generate_tts.py --dry-run` measured it at 197s / 5971 frames, confirming the 3-7min auto-mode target. That is a clear, concrete improvement over the baseline. CONFIRMED BUG: `scripts/tts/sections.py` uses `r'\[SECTION:(\w+)\]'`, which does not match hyphens, so the skill's own shipped `templates/podcast_en.txt` markers `[SECTION:content-1..3]` are silently dropped — my first dry-run reported "Detected 5 sections" and folded three content blocks into `overview`; renaming to `content_1` gave the correct "Detected 8 sections". SKILL.md also points to `CLAUDE.md` ("full command reference") and `examples/`, both HTTP 404 in this copy. No security smells: no curl|sh, no base64 blobs, no secret exfiltration, paths use `${SKILL_DIR}`. outputMeasured=false because the advertised deliverable — a 4K MP4 — needs a scaffolded Remotion project, npm install and an ffmpeg render I did not stand up; only the scripting phase was measured. Note this is a vendored copy of upstream Agents365-ai/video-podcast-maker inside a Chinese local-search app repo.
Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works
- Installs cleanly 4/5
- Triggers reliably 5/5
- Output vs. baseline 6/10
- Docs & honesty 3/5
What Video Podcast Maker does
A 15-step pipeline that turns a topic into a finished video podcast: web research, a sectioned narration script, TTS audio with timing.json and SRT, a Remotion composition rendered at 3840x2160, FFmpeg background-music mixing, thumbnails, and platform-specific publish metadata. Triggers when you ask for a video podcast, explainer video, or Bilibili/YouTube knowledge video on a given topic, or ask it to learn visual style from a reference video or image. Supports seven TTS backends (Edge TTS free by default, plus Azure, Doubao, CosyVoice, ElevenLabs, OpenAI, Google) and zh-CN/en-US output with Chinese polyphone correction.
How to install Video Podcast Maker
git clone https://github.com/dtsola/xiaoyaosearch /tmp/xiaoyaosearch
cd /tmp/xiaoyaosearch
mkdir -p ~/.claude/skills
cp -R /tmp/xiaoyaosearch/.claude/skills/video-podcast-maker ~/.claude/skills/
cp -R /tmp/xiaoyaosearch/.claude/skills/remotion-best-practices ~/.claude/skills/ # hard dependency
brew install ffmpeg node # macOS; node + npx + ffmpeg + python3 required
pip install -r ~/.claude/skills/video-podcast-maker/requirements.txt
python3 ~/.claude/skills/video-podcast-maker/scripts/check_prereqs.py # expect: ALL_OK (backend=edge)
# You also need an existing Remotion project; videos land in videos/{name}/ inside it.
# Optional: set AZURE_SPEECH_KEY / OPENAI_API_KEY / etc. only if you switch off the free `edge` backend.
Skills live in ~/.claude/skills/ (global) or .claude/skills/
(per-project). Restart Claude Code after installing.
Commands — how to trigger Video Podcast Maker
-
/video-podcast-makerTopic to 4K video podcast: research, script, TTS, Remotion render, BGM mix
It also activates on plain-language prompts like these:
-
Make a video podcast about this topic -
Learn this reference video's visual style and apply it -
Generate a video podcast with background music
Frequently asked questions
- Is the Video Podcast Maker skill free?
- Yes. The skill itself is free from dtsola/xiaoyaosearch. SkillProof publishes the install command and an independent test verdict at no cost.
- Does Video Podcast Maker work with Claude Code?
- We tested it with Claude Code 2.x (agent harness) on Jul 21, 2026. Verdict: Works with setup. I downloaded the skill's real scripts and ran them. `check_prereqs.py` printed `ALL_OK (backend=edge)` and `migrate_prefs.py --dry-run` printed `[dry-run] Would create user_prefs.json at v1.6 from template`, so the Python side genuinely works. For output I wrote a baseline "why HTTP/3 uses QUIC" script before reading the body (4 scenes, 60s, ad-hoc markdown, no machine-readable format), then followed references/workflow-script.md Steps 1-4 and got topic_definition.md plus a podcast.txt with 8 `[SECTION:xxx]` blocks, density tiers mapped to the named component library (ComparisonCard/FlowChart/DataBar), and the platform+language outro from the skill's CTA table; the skill's own `generate_tts.py --dry-run` measured it at 197s / 5971 frames, confirming the 3-7min auto-mode target. That is a clear, concrete improvement over the baseline. CONFIRMED BUG: `scripts/tts/sections.py` uses `r'\[SECTION:(\w+)\]'`, which does not match hyphens, so the skill's own shipped `templates/podcast_en.txt` markers `[SECTION:content-1..3]` are silently dropped — my first dry-run reported "Detected 5 sections" and folded three content blocks into `overview`; renaming to `content_1` gave the correct "Detected 8 sections". SKILL.md also points to `CLAUDE.md` ("full command reference") and `examples/`, both HTTP 404 in this copy. No security smells: no curl|sh, no base64 blobs, no secret exfiltration, paths use `${SKILL_DIR}`. outputMeasured=false because the advertised deliverable — a 4K MP4 — needs a scaffolded Remotion project, npm install and an ffmpeg render I did not stand up; only the scripting phase was measured. Note this is a vendored copy of upstream Agents365-ai/video-podcast-maker inside a Chinese local-search app repo.
- What is the Video Podcast Maker SkillProof Score?
- 7.2/10 — installs cleanly 4/5, triggers reliably 5/5, output vs. baseline 6/10, docs & honesty 3/5.
- How do I install Video Podcast Maker?
- Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.
- Can I use Video Podcast Maker with Cursor, Copilot, Gemini CLI, Codex or other AI tools?
- The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.