lookalike-contentlisted

Analyse a data dump of posts to find patterns in what's working, create a winning content profile, then generate 10 content ideas that match those patterns using external research. Use this skill when the user wants to find content patterns, analyse what's working in their content, reverse-engineer successful posts, find lookalike content ideas, or generate ideas based on proven patterns. Also trigger when the user mentions 'what's working in my content', 'find patterns', 'lookalike content', 'content ideas based on what works', 'analyse my posts', or 'reverse engineer my best content'.
t0ddharris/mktg-os · ★ 0 · Data & Documents · score 58

Install: claude install-skill t0ddharris/mktg-os

# Lookalike Content ## Overview This skill takes a data dump of posts from any platform, analyses them for patterns in what makes the content successful, creates a winning content profile, then uses that profile to research and generate 10 content ideas that follow the same patterns. One skill, one run, three phases: 1. **Analyse** — read the data dump, find patterns in structure, topics, hooks, emotional register, format 2. **Profile** — create a winning content DNA document 3. **Generate** — research trending topics and produce 10 ideas that match the winning patterns ## Prerequisites ### API Key | Variable | Purpose | |---|---| | `TAVILY_API_KEY` | Researching trending topics for content idea generation | ## Context Loading — Mandatory Before running, read these files: 1. `/brain/positioning-and-messaging.md` — brand voice, messaging, words we use/avoid 2. `/brain/truth.md` — verified product and company facts ## Workflow ### Step 1: Get the Data Dump The user provides files containing their posts. These come in messy formats — the skill handles the conversion. Accepted formats: - A folder of HTML files (e.g., Substack export) - A CSV export (from analytics tools, newsletter platforms, etc.) - A JSON archive (e.g., X data export) - A markdown or text file with posts listed - Multiple files uploaded at once - Raw text pasted directly - A path to files anywhere on disk ``` Upload your content data dump. This can be: - A folder of HTML files (Substack export)