What problem does it solve? Most changelogs are either vague summaries written from memory or unreadable commit dumps. This Skill turns changelog writing into an evidence-driven research process, producing an agent-friendly historical record grounded in real git history, release metadata, and issue tracker data. ## Core Features & Use Cases - Evidence-first research workflow: Builds a version spine from git tags and GitHub Releases, distinguishes releases from plain tags, and mines issue trackers (beads, GitHub Issues, Linear, Jira, milestone docs) for intent. - Chunked reconstruction for large histories: Splits long histories into bounded slices with a durable research memo and coverage ledger, updating CHANGELOG.md incrementally after each chunk. - Automation scripts and templates: Includes scripts to bootstrap the workspace, build a version timeline, extract tracker workstreams, cluster commits into capability waves, and validate the finished changelog, plus copy-paste markdown templates. - Use Case: Point it at a multi-year repository and ask for a full-history CHANGELOG.md; it researches tags, releases, commits, and tracker items chunk by chunk, then produces a version timeline plus thematic capability sections with live commit links. ## Quick Start Ask the AI to rebuild this repository's CHANGELOG.md from the full git history, tags, releases, and issue tracker, using chunked research if the history is large.