Turn Any Link into a Competitor Teardown Report

Where this fits

A creator asks your agent: "Break down this competitor's account — why does every video hit 100k views?" You have a link, but no time to manually watch, transcribe, and map out scenes. This skill turns that link into a structured teardown report: a reusable viral formula, a timestamped scene-by-scene breakdown of visuals and voiceover, and a positioning analysis. It's built for agents that need to deliver competitive intelligence without burning hours on manual research.

Why agents benefit

• Auto-detects platform and granularity: Pass a single video or an entire account link from Douyin, Xiaohongshu, Bilibili, YouTube, or WeChat Official Account. The skill figures out what to fetch and how deep to go.
• Goes beyond captions: Downloads short videos, transcribes the voiceover with Whisper, and extracts a contact sheet of visual frames for multimodal analysis. You get the actual content, not just the text.
• Outputs a fixed, reusable template: Every teardown follows the same three-part structure (formula, scene breakdown, positioning), making it easy to compare competitors side-by-side.
• Graceful fallback: If content can't be fetched, it asks the user to paste the transcript or caption instead of failing with an error — keeps the agent flow alive.
• Auto-archives results: Saves all reports to benchmarks/ so you build a growing library of competitive insights without extra prompting.

Practical scenarios

Analyzing a single viral video

A user drops a YouTube link and asks "Why did this get 2M views?" The agent runs the skill, downloads the video, transcribes the speech, extracts key frames, and produces a scene-by-scene breakdown with timestamps, hooks, and pacing notes. The user gets a clear formula they can apply to their next piece.

Full account teardown for positioning

"Analyze this Xiaohongshu account's content strategy." The skill pulls multiple posts, identifies recurring visual symbols, content pillars, and the persona's voice. It outputs a positioning report that highlights what makes the account distinct and how to differentiate.

Batch competitor research for content planning

An agent managing a content calendar runs teardowns on three competitor accounts weekly. Each run appends to benchmarks/, and the agent can later query that folder to spot trends in formats, hooks, and CTAs across the niche.

Add it to your agent workflow

Install with a single command:

npx skills add ttfake92-lab/skills --skill yyl-benchmark-breakdown

Then trigger it with natural language. For example:

User: teardown https://v.douyin.com/abc123/
Agent: (runs dependency check, fetches content, outputs report)

The skill handles the dependency check automatically (Docker for download APIs, ffmpeg/whisper for transcription) and will guide the user to fix missing tools if needed — no dead ends.

Compare before adopting

Before committing, compare this skill against alternatives on four axes: quality signals — does the transcription and frame extraction actually work on your target platforms? maintenance freshness — check the GitHub repo for recent commits and issues; platform APIs change fast. alternatives — consider manual analysis or simpler scraping scripts if your needs are narrow. workflow fit — does the three-part output template match how your agent presents results, or do you need to adapt it?

Why it is worth tracking

With 206 stars and an active author, this skill shows real community traction. It solves a pain point that many content agents overlook — deep, multimodal teardowns instead of surface-level metadata. It's worth evaluating if you build agents for creators, marketers, or competitive research. For a deeper look, see the skill page and decide if it earns a permanent spot in your agent's toolkit.