Architecture
Does the architecture still say what the code does?
Architecture erodes one pragmatic shortcut at a time, and by the time it is visible it is expensive. This package runs the complete assessment from Robert C. Martin's Clean Architecture against a codebase: twenty-nine chapters, from the SOLID principles through component cohesion and coupling to where the boundaries actually sit, each chapter's checklist items verified one step at a time.
The result is not an opinion about the architecture. It is a list of specific violations, each with a file and line, a severity, an effort estimate and a ticket-ready recommendation — and a cross-cutting audit at the end that checks the findings against each other.
Architecture Basic
Free
Pro and up · included with your license
Teams deciding whether a system can take the next feature or needs a boundary first, and anyone who has inherited a codebase whose architecture diagram no longer matches the imports.
What runs
The workflows inside
Clean Architecture Assessment
Orchestrates twenty-nine chapter workflows as child runs — the SOLID principles, the component principles, the architecture chapters from What Is Architecture through The Test Boundary, the details chapters on databases, the web and frameworks — each checking every checklist item in its own step and appending reviewed findings to a shared log.
Clean Architecture Cross-Cutting Audit
A final six-item quick audit that gathers evidence with file and line citations, marks each item violated, passed or not assessable, and assesses severity, effort and a recommendation per violation.
Clean Architecture Compile Report
Dedupes the findings across all chapters, ranks them, verifies each recommendation and writes the report behind a final deliverable review.
How a run goes
Step by step, in the order the engine enforces
- 1
Start from the repository root
The agent starts the assessment with the target directory and where to write the report. Every chapter runs as its own child instance, so a large codebase does not overwhelm one context.
- 2
Every checklist item is a step
The engine hands over one item at a time. The agent gathers evidence, decides violated, passed or not assessable, and records it. It cannot mark a chapter done with items unchecked.
- 3
Findings are reviewed, then logged
Each chapter's findings loop through review until they pass, then are appended to the shared log. Only that chapter's findings ever pass through a prompt; the accumulated log stays on disk.
- 4
The report is compiled last
Deduped, ranked and verified across all twenty-nine chapters, with the cross-cutting audit's verdicts alongside.
What you get
At the end of a run
- ✓A Clean Architecture report with findings per chapter and a ranked remediation list
- ✓A cross-cutting audit marking each of six structural items violated, passed or not assessable
- ✓The findings log, with a file and line for every entry
Read next
Related packages
- Code Quality →
Run a full ISO/IEC 5055 and Clean Code assessment of any codebase with your coding agent, every check its own verified step. Free with every CodyCody plan.
- Boy Scout Coding →
A code change by your coding agent: scoped, designed, approved at two human gates, reviewed from three angles, covered by tests — and the code left cleaner.