What problem does it solve? Removing a shipped feature is riskier than it looks: half-removed concepts leave behind tables, routes, published types, and agent assets that still instruct users to use the retired capability. This Skill turns feature removal into a measured, classified, and provable process so nothing stale survives the release. ## Core Features & Use Cases - Residue Inventory: Run calibrated grep-based inventories from inside each repository root, with exclusion patterns that correctly handle root-level build directories, and record per-extension breakdowns before editing anything. - Three-Class Classification: Sort every hit into live code (deleted with reference scans), shipped documentation and agent assets (updated or deleted), and historical records (never rewritten), with rules for detecting misclassified derived artifacts. - Gated Irreversible Changes: Handle schema drops and data migrations on a separate approval track, sequenced after the code that reads them is gone. - Republish Cascade and Zero-Residue Proof: Bump and republish dependent workspace packages, then prove completion by re-running the opening inventory plus a reference scan for every deleted file and script name at the final commit. - Use Case: When retiring a domain capability from a multi-repository product, follow this flow to produce a tracking record with before/after inventory numbers, class breakdowns, schema decisions, and a closing record a reviewer can independently verify. ## Quick Start Use the feature-retirement-flow skill to plan and execute the removal of the retired feature from this repository set, starting with a residue inventory.
