neumie Improve Codebase Architecture
Surface architectural friction and propose deepening opportunities: refactors that turn shallow modules into deep ones. The aim is testability and AI-navigability.
Follow the codebase-design skill for design vocabulary and principles. Follow the domain-model skill when maintaining CONTEXT.md or ADRs.
Process
1. Explore
Scope the scan before exploring. Deepening pays off when future changes land in that area, so avoid speculative whole-repo review:
- If the user names a module, subsystem, or pain point, focus there.
- Otherwise inspect a useful stretch of
git log --onelinefor recurring files and areas. Start with those hot spots; widen only when history is scattered.
Read CONTEXT.md if present and use its vocabulary throughout. Read relevant ADRs under docs/adr/ before exploring.
Then use a read-only exploration sub-agent to walk the scoped codebase when available; otherwise explore directly. Don't follow rigid heuristics; note where you experience friction:
- Where does understanding one concept require bouncing between many small modules?
- Where are modules shallow — interface nearly as complex as the implementation?
- Where have pure functions been extracted just for testability, but the real bugs hide in how they're called (no locality)?
- Where do tightly-coupled modules leak across their seams?
- Which parts of the codebase are untested, or hard to test through their current interface?
Apply the deletion test to anything you suspect is shallow: would deleting it concentrate complexity, or just move it? A "yes, concentrates" is the signal you want.
2. Present candidates as an HTML report
Write a self-contained HTML file to the OS temp directory so nothing lands in the repo. Resolve the temp dir from $TMPDIR, falling back to /tmp (or %TEMP% on Windows), and write to <tmpdir>/architecture-review-<timestamp>.html so each run gets a fresh file. Open it for the user — xdg-open <path> on Linux, open <path> on macOS, start <path> on Windows — and tell them the absolute path.
The report uses Tailwind via CDN for layout and Mermaid via CDN for diagrams where a graph, flow, or sequence reliably communicates the structure. Mix Mermaid with hand-crafted CSS/SVG visuals: use Mermaid when relationships are graph-shaped, and hand-built divs/SVG for mass diagrams, cross-sections, and collapse diagrams. Each candidate gets a before/after visualization.
See HTML-REPORT.md for the full HTML scaffold, diagram patterns, and styling guidance.
For each candidate, render a card with:
- Files — which files/modules are involved
- Problem — why the current architecture is causing friction
- Solution — plain English description of what would change
- Benefits — explained in terms of locality and leverage, and also in how tests would improve
- Before / After diagram — side-by-side, custom-drawn, illustrating the shallowness and the deepening
- Recommendation strength — one of
Strong,Worth exploring,Speculative, rendered as a badge
End the report with a Top recommendation section: which candidate you'd tackle first and why.
Use CONTEXT.md vocabulary for the domain, and codebase-design vocabulary for architecture.
ADR conflicts: if a candidate contradicts an existing ADR, only surface it when the friction is real enough to warrant revisiting the ADR. Mark it clearly (e.g. "contradicts ADR-0007 — but worth reopening because…").
Do NOT propose interfaces yet. After the file is written, ask the user: "Which of these would you like to explore?"
3. Grilling loop
Once the user picks a candidate, follow the grilling skill to resolve constraints, dependencies, the shape of the deepened module, what sits behind the seam, and which tests survive.
Follow the domain-model skill as naming and design decisions crystallize; it owns glossary updates and ADR eligibility. For a rejected candidate, offer an ADR only when the reason is load-bearing for future explorers, not merely "not worth it right now."
To explore alternative interfaces, follow the codebase-design skill and its design-it-twice reference.