cs-audit

The final acceptance stage of the CodeStable feature workflow that verifies implementation against the design document, updates architecture and requirement docs, closes roadmap items, and produces an acceptance report as the closure record.

Stage 1 of the codestable feature workflow that drafts a comprehensive design document (covering architecture, interface contracts, implementation hints, and test design) plus an action checklist, which serve as the single source of truth for the subsequent implementation and acceptance stages.

A CodeStable skill that serves as the entry point for exploratory discussions — it diagnoses vague ideas through one or two rounds of dialogue and routes them to the right downstream: directly to feature design, into a feature brainstorm note, or to roadmap decomposition for large requirements.

A skill that onboards a repository into the CodeStable workflow system, either scaffolding a fresh directory structure from scratch or auditing and migrating existing scattered documents into the standard codestable/ layout.

A code exploration skill that turns the "question → read code → conclusion" process into a searchable, evidence-based document, supporting three exploration types: question (code investigation around a specific issue), module-overview (structure/boundary mapping), and spike (lightweight multi-directional technical probe).

Stage 2 of the codestable feature workflow that translates an approved design document into code by following the checklist step-by-step, enforcing three core disciplines: write the minimum necessary code, only modify intended files, and never make decisions not covered by the design document.

A CodeStable skill that serves as a one-stop architecture center — draft new architecture documents, refresh existing ones, or run architecture health checks with three modes: new, update, and check.

A knowledge-capture skill in the codestable workflow that distills engineering lessons into searchable learning documents — pitfall track records bugs, traps, and failed attempts while knowledge track records best practices and reusable patterns, so teams avoid repeating the same mistakes.

A CodeStable skill for behavior-equivalent code improvements — it routes refactoring work through a scan-design-apply pipeline with human checkpoints at each stage, ensuring AI only touches what it can do reliably without changing external behavior.

A CodeStable skill that routes new feature development through a four-stage workflow — brainstorm, design, implementation, and acceptance — inserting a design document between requirements and code to prevent scope drift and leave behind a traceable record.