What problem does it solve?
Dependency upgrades often break builds, introduce silent regressions, or leave lockfiles inconsistent. This Skill upgrades dependencies in small, attributable batches with breaking-change research and per-batch verification, so failures can be isolated and reverted without touching unrelated work.
Core Features & Use Cases
- Batched upgrades with rollback: Groups updates by prerequisite order (runtime, build tooling, frameworks, leaf packages), verifies each batch, and reverts only the failed batch.
- Breaking-change and security research: Checks official release notes, migration guides, and advisories before applying consequential updates, and re-runs security audits after each batch.
- Evidence-based reporting: Produces a verdict (UPDATED, PARTIAL, NO_CHANGE, BLOCKED) with a batch table, advisory status, and verification evidence.
- Use Case: A repository has 30 outdated npm packages including one major framework upgrade. The Skill inventories updates, researches the framework's migration guide, applies routine patches in small batches, migrates code for the major version, and reverts any batch that fails tests.
Quick Start
Ask the AI to upgrade the outdated dependencies in this repository in safe verified batches, researching breaking changes first and reverting anything that fails the build or tests.