Social media research is noisy and multi-platform: useful signals live across short videos, comments, ad libraries and community threads. This repository reframes that problem by shipping ready-made "skills" an AI agent can call to gather, normalize and transform public social data into business-ready artifacts — not just raw scrapes.
What Sets It Apart
- Workflow-first, not endpoint-first — the repo provides end-to-end research workflows (outlier-post-finder, comment-mining, transcript-intelligence, ad-library-teardown, trend-discovery, content-repurposing) so agents deliver practical outputs (tables, briefs, swipe files) instead of raw JSON. So what: saves prompt engineering and glue-code work when automating social research tasks.
- Agent-native integrations — designed to run with agent runtimes like Claude Code, Cursor, OpenAI Codex, Gemini CLI and others. So what: you can plug these skills into different agent environments without redesigning the research logic.
- Public-data and repeatability focus — emphasizes baseline-aware analysis, exact language preservation, and citation of source URLs. So what: reports are reproducible, auditable, and tuned to a creator’s own performance rather than vanity metrics.
- Practical output templates — each skill produces artifacts (VOC reports, outlier tables, competitor briefs, content ideas) intended to be immediately actionable for content teams, growth, product demand validation, or influencer outreach.
Who it's for + Tradeoffs
Great fit if you need automated, repeatable social research pipelines that agents can run at scale—content strategists, growth teams, product researchers, or engineers building agent-driven workflows. It’s especially useful when you want cited examples, exact audience language, and production-ready artifacts rather than raw data. Look elsewhere if you need private/authenticated data (this repo and its data layer work only with public social data), deep platform-native analytics behind logins, or a turnkey SaaS dashboard — this is a toolkit for agents and engineers, not a hosted analytics product.