What problem does it solve? AI-generated writing never sounds like you because the model has never seen your actual sentences. This Skill mines your real writing from two sources and produces a calibrated voice skill that drafts content matching your authentic style. ## Core Features & Use Cases - Dual-source corpus building: Mines Littlebird memory via MCP (captured writing, messages, meeting speech) and ingests a Facebook data export (posts, messages, profile) with a screenshot-guided export walkthrough. - Scripted stylometric analysis: Counts dashes, ellipses, emoji, exclamation rates, hashtags, openers, and length distribution, plus qualitative register reading across both corpora. - User-confirmed output: Every biography fact and sample draft is confirmed via approval checkpoints before packaging into a skill with fingerprint, anti-AI rules, corpus, and samples references. - Use Case: A founder wants Claude to draft social posts and replies that sound like them. They run this Skill, export their Facebook data while Littlebird is mined in parallel, approve the analysis and samples, and receive an installed voice skill that other skills compose with. ## Quick Start Ask Claude to build my voice skill using both my Littlebird data and my Facebook export.