What problem does it solve? AI-drafted personal messages often sound generic, carry recognizable machine-writing patterns (em dashes, buzzwords, clipped fragments), and ignore how the actual sender writes. This Skill drafts person-to-person communication in the sender's own voice and removes detectable AI tells before the text is sent. ## Core Features & Use Cases - Voice-first drafting: Prioritizes the sender's own instructions, personal style files, and real writing samples over generic defaults, and explicitly avoids applying organizational brand voice to personal messages. - AI-tell removal: Strips mechanical tells (em dashes, always-replace words like "delve" and "leverage", empty intensifiers, copy-paste fingerprints) and structural tells (false agency, antithesis patterns, clipped fragments) via a bundled Python linter. - Reusable voice profiles: Captures a sender's confirmed preferences into a personal skill or profile using the included voice-profile template. - Use Case: Ask the assistant to draft a Slack reply to a colleague about a delayed deliverable; it writes in your established tone, runs lint.py to catch AI-sounding phrasing, and returns a message that reads like you wrote it. ## Quick Start Draft a short email to my manager asking to move Thursday's meeting to next week, written in my usual direct and informal style.
