What problem does it solve? Teams often miss impactful technical improvements because they only watch dependency updates, while real gains hide in slow CI pipelines, inconsistent code style, untested branches, and drift between conventions and actual practice. This Skill investigates the current repository from multiple angles and proposes only improvements backed by primary sources and real measurements. ## Core Features & Use Cases - Multi-angle investigation: Examines dependencies, CI duration and failure trends, notation inconsistencies, and test coverage gaps, optionally scoped via an argument (deps / ci / style / test). - Evidence-based proposals: Cross-checks claims against primary sources (GitHub Releases, RFCs, official blogs) and requires measured before/after numbers for any performance claim. - Strict filtering: Drops proposals that violate project conventions, duplicate existing Issues/PRs, or address non-problems, and caps output at three proposals. - Use Case: Run it in a monorepo before a quarterly maintenance cycle to discover that a CI workflow's worst-case runtime doubled, a lint rule could replace manual style fixes, and a dependency deprecation can be handled cheaply now. ## Quick Start Ask the AI to run /propose-tech-updates in your repository to investigate dependencies, CI, style, and tests and return up to three evidence-backed improvement proposals.
