What problem does it solve? Engineering teams often lack a structured way to identify what to fix or build next in a codebase. This Skill performs a read-only audit of a repository, gathers file:line evidence, vets findings by impact and effort, and routes each selected suggestion into the repository's canonical OpenSpec and Beads workflow without inventing parallel backlogs. ## Core Features & Use Cases - Read-only codebase audits: Covers correctness, security, performance, tests, architecture, dependencies, developer experience, and documentation at quick, standard, or deep effort levels. - Scoped reviews: Supports domain audits (security, performance, tests), branch reviews labeling findings as introduced or pre-existing, roadmap direction analysis, and plan or proposal reviews. - Self-contained suggestion contracts: Every selected finding includes evidence, definition of done, blast radius, and exactly one route (attach, feature, ad-hoc task, or research/decision map) so a fresh executor can act without session context. - Use Case: Ask for a security audit of your repository before a release; the advisor inspects input boundaries, secret handling, and authz checks, then returns a prioritized list of verified findings with an ordered execution queue. ## Quick Start Ask the agent to audit this repository and tell you what to improve next, optionally scoping it to a domain like security or a branch review.
