Communitygithub.com

shengdabai/Tony-Claude-Code-Skills

Accepts short video links or shared captions from Douyin, Xiaohongshu, WeChat Video Channel, etc., queries video/account data, extracts audio transcripts, and archives them as Markdown files organized by author and title. Use this when you need a complete breakdown of a short video, including data and transcript. Before running, it checks TikHub and Qingdou credentials, and if missing, guides new users through purchase and secure configuration in an understandable way.

What is Tony-Claude-Code-Skills?

This skill, named 'dbs-video-extract: Short Video Info Extraction', provides a unified entry point for processing shared short video captions. It leverages TikHub to fetch structured data from platforms like Douyin, Xiaohongshu, and WeChat Video Channel, and Qingdou to extract speech-to-text transcripts. The skill outputs a data summary and the path to a Markdown file. It supports Claude Code and similar AI agents, requiring API keys for both services. The workflow includes a credential pre-check via `--check-keys`, guiding users through purchasing and securely configuring keys if missing. The skill is designed to handle partial results (only data or only transcript) if one service is unavailable, and it clearly communicates the capabilities and limitations to the user. Default mode is `both`, but can be switched to `data` or `transcript` based on user preference or available credentials. It prioritizes reading input from stdin to avoid shell interpretation issues, and saves transcripts locally in a structured directory. The skill also includes detailed instructions for handling missing credentials, including legal disclaimers and step-by-step purchase guidance, making it accessible even for non-technical users.

Works with✓Claude Code~Codex CLI~Cursor
npx skills add https://github.com/shengdabai/Tony-Claude-Code-Skills/tree/HEAD/skills/dbs-video-extract

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

dbs-video-extract: Short Video Info Extraction

Goal: Process a single short video's shared caption from one entry point. TikHub provides structured data for Douyin, Xiaohongshu, and WeChat Video Channel; Qingdou provides audio transcripts for short videos; the final output is a data summary and the path to the Markdown transcript file.

Check Credentials First

Before any extraction task, run a read-only pre-check:

python3 scripts/extract_video.py --check-keys

Handle based on credential_state:

StateCapabilityMust Tell User
completeBoth data and transcript are availableBoth parts can be executed
data_onlyOnly TikHub data queries are availableCan only retrieve video/account data, cannot generate audio transcript
transcript_onlyOnly Qingdou transcript is availableCan only retrieve audio transcript, cannot query TikHub video/account data
unavailableNeither part is availableCurrently unavailable; first explain credentials, then guide the user through purchase and configuration

Proceed if only one key is present; do not require the user to obtain the other key. Before execution, clearly state which part can be handled this time, and default to narrowing the mode to the available side. If the user still insists on both, keep the results from the available side and mark the other side as missing credentials.

Stop extraction if both keys are missing. Do not conclude with terms like "API Key" or "interface authentication" directly. First use this explanation:

To make this Skill work, you first need to activate external data services. An API Key is a credential string provided by the service after activation—think of it as a special pass for this Skill to call the service. You don't need to understand the technical details, and you should not share it in the chat.

Then tell the user: "Currently, neither service has been activated. Without purchasing and configuring at least one, this Skill cannot be used." Then follow references/api-setup.md to complete requirement selection, purchase, credential saving, and connection testing.

Do not make payments on behalf of the user, accept service agreements, or handle complete keys. Do not ask the user to send the key into the conversation. Purchase links, step-by-step guidance, secure configuration scripts, and connection tests are available in references/api-setup.md.

Whenever showing a purchase or top-up address to the user, you must first display the following notice exactly as written, without omission or modification:

dbskill is a free, open-source project and has no affiliation, partnership, or financial interest with TikHub or Qingdou. Top-ups are entirely at the user's discretion; all related transactions, services, disputes, and risks are the user's sole responsibility. Unless otherwise provided by law, dbskill assumes no liability.

Do not repeat this notice when the user has not entered a purchase or top-up flow.

Dialogue Requirements When Credentials Are Missing

When any service is missing, the reply must form the following continuous path:

  1. First, explain the API Key in plain language;
  2. Ask the user to select based on their needs: full functionality, data only, or transcript only;
  3. Explain which service each choice requires purchasing, and what will be missing if only one is bought;
  4. Display the fixed legal notice, then show the corresponding top-up address;
  5. Provide step-by-step instructions for registration/login, viewing page plans, topping up, entering the personal center or key management page, and obtaining credentials;
  6. Tell the user that after obtaining the credentials, they should not send them in the chat, but reply "TikHub topped up," "Qingdou topped up," or "Both topped up";
  7. After the user confirms top-up, guide them to run the secure configuration script in their local terminal:
python3 scripts/configure_api_key.py tikhub
python3 scripts/configure_api_key.py qingdou

The script will read the credentials discreetly in the terminal and save them to a local private configuration; do not display the full credentials in commands, replies, or logs. After saving, re-run --check-keys until the corresponding state becomes available.

Do not send links all at once, and do not assume the user knows where to find credentials. If the service provider's page names change, guide using candidate locations like "Personal Center," "API Key," "Key Management," or "Developer Settings". If still not found, ask the user to send a screenshot of the page with account, balance, and credentials obscured, then continue troubleshooting.

When the user only invokes this Skill without providing a link or specifying what kind of result they need, after completing the credential pre-check, do only two things: explain the API Key, and ask the user to reply 1, 2, or 3 to choose full functionality, data only, or transcript only. After receiving their choice, show the legal notice, corresponding top-up address, and purchase steps. If the user has already sent a video link or explicitly requested a full extraction, default to full functionality without asking again.

Default Execution

Prefer standard input to avoid shell interpretation of special characters in shared captions:

python3 scripts/extract_video.py \
  --output-dir "/absolute/path/ShortVideoTranscripts" \
  --stdin

Default mode is both:

  1. Extract the link from the full shared caption;
  2. Call the TikHub MCP based on the link platform: Douyin supports user homepage or individual video, Xiaohongshu and Video Channel support individual video;
  3. Submit the full shared caption to the Qingdou API, polling for the transcript;
  4. Save the transcript as {output_dir}/{author}/{title}.md;
  5. Output a JSON summary containing credential status, TikHub data summary, and transcript file path.

If either TikHub or Qingdou fails, continue with the other side and mark partial_success in the summary. Do not discard results already obtained due to a single-side failure.

Optional Modes

Switch only when the user explicitly requests one side, or the pre-check confirms only one side's key is available:

python3 scripts/extract_video.py --mode data --stdin
python3 scripts/extract_video.py --mode transcript --stdin
  • data: Query only TikHub data, supports Douyin, Xiaohongshu, and WeChat Video Channel;
  • transcript: Extract only the transcript, supports platforms that the Qingdou API can parse;
  • both: Execute both sides simultaneously.

Skip transcript writing by default if the same source already exists to protect user edits; add --overwrite when the user explicitly requests regeneration. Add --raw-data when the full TikHub response is needed.

Keys and Costs

The script reads keys in the following order:

  1. Environment variables: TIKHUB_API_KEY, QINGDOU_API_KEY;
  2. User-specified local files: TIKHUB_API_KEYS_FILE, QINGDOU_API_KEYS_FILE;
  3. ~/.config/dbs/API_Keys.md;
  4. macOS Keychain: dbs-tikhub-api-key, dbs-qingdou-api-key.

Do not write keys into the Skill, command arguments, Markdown, Git, or logs. TikHub queries and Qingdou transcriptions may incur costs; inform the user before execution. No need to confirm again if the user has already explicitly requested extraction.

Data Boundaries

  • TikHub handles video/account profiles, statistics, and media metadata; prioritize App V3 for Douyin videos, fall back to Web only once if no valid data; use App V2 video note details for Xiaohongshu; use Channels V2 video details for Video Channel.
  • Qingdou handles audio transcripts. Faithfully preserve the API response body; do not supplement, polish, or correct verbal errors.
  • Current TikHub-specific data parsing supports Douyin, Xiaohongshu, and WeChat Video Channel; other platforms recognizable by Qingdou can still perform transcript-only extraction.
  • Public, deleted, private, copyright, and visibility restrictions are determined by the API response; do not attempt to bypass them.
  • Do not automatically download media, publish content, commit to Git, or push.

See references/tikhub-api.md for TikHub call details; see references/qingdou-api.md for Qingdou status codes and compatibility logic. Only read these when troubleshooting or maintaining the corresponding part.

Deliverables

After completion, provide a concise report:

  • Which API key is currently configured and the scope of available capabilities;
  • Whether the data query was successful, which MCP tool was used, and whether an App→Web fallback occurred;
  • Key video/account data;
  • Whether the transcript was successfully generated or skipped, and the absolute path to the Markdown file;
  • Specific reasons for missing keys or execution failure on one side;
  • Whether the interface explicitly indicated that a charge was applied.

Do not display API keys, batchId, or unnecessary internal request identifiers in the reply.

After completing the current task, end directly. Only if the user explicitly asks about the next step, and the /dbs environment is already installed, briefly prompt: "If you're unsure about the next step, you can enter /dbs."

Individual skills in this repo

This repo contains 8 individual skills — each has its own dedicated page.

shengdabai/Tony-Claude-Code-Skills

Downloads videos from YouTube and other platforms for offline viewing, editing, or archival. Handles various formats and quality options.

shengdabai/Tony-Claude-Code-Skills

Read-only queue dashboard for workspace-aware ship. (gstack)

shengdabai/Tony-Claude-Code-Skills

Anime.js adapter patterns for HyperFrames. Use when writing Anime.js animations or timelines inside HyperFrames compositions, registering animations on window.__hfAnime, making Anime.js seek-driven and deterministic, or translating Anime.js examples into render-safe HyperFrames HTML.

shengdabai/Tony-Claude-Code-Skills

Author a new HyperFrames registry block (caption style, VFX block, transition, lower third) or component (text effect, overlay, snippet) and ship it as an upstream PR to the hyperframes repo. Use ONLY when the user wants to CONTRIBUTE to the public catalog — for in-project caption/transition authoring use the `hyperframes` skill, for installing existing registry items use the `hyperframes-registry` skill.

shengdabai/Tony-Claude-Code-Skills

CSS animation adapter patterns for HyperFrames. Use when authoring CSS keyframes, animation-delay based timing, animation-fill-mode, animation-play-state, or CSS-only motion that HyperFrames must seek deterministically during preview and rendering.

shengdabai/Tony-Claude-Code-Skills

GSAP animation reference for HyperFrames. Covers gsap.to(), from(), fromTo(), easing, stagger, defaults, timelines (gsap.timeline(), position parameter, labels, nesting, playback), and performance (transforms, will-change, quickTo). Use when writing GSAP animations in HyperFrames compositions.

shengdabai/Tony-Claude-Code-Skills

HyperFrames CLI dev loop — `npx hyperframes` for scaffolding (init), validation (lint, inspect), preview, render, and environment troubleshooting (doctor, browser, info, upgrade). Use when running any of these commands or troubleshooting the HyperFrames build/render environment. For asset preprocessing commands (`tts`, `transcribe`, `remove-background`), invoke the `hyperframes-media` skill instead.

shengdabai/Tony-Claude-Code-Skills

Asset preprocessing for HyperFrames compositions — text-to-speech narration (Kokoro), audio/video transcription (Whisper), and background removal for transparent overlays (u2net). Use when generating voiceover from text, transcribing speech for captions, removing the background from a video or image to use as a transparent overlay, choosing a TTS voice or whisper model, or chaining these (TTS → transcribe → captions). Each command downloads its own model on first run.

Related Skills