improve-codebase-architecture

# 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.

## Glossary

Use these terms exactly in every suggestion. Consistent language is the point — don't drift into "component," "service," "API," or "boundary." Full definitions in [LANGUAGE.md](LANGUAGE.md).

- **Module** — anything with an interface and an implementation (function, class, package, slice).
- **Interface** — everything a caller must know to use the module: types, invariants, error modes, ordering, config. Not just the type signature.
- **Implementation** — the code inside.
- **Depth** — leverage at the interface: a lot of behaviour behind a small interface. **Deep** = high leverage. **Shallow** = interface nearly as complex as the implementation.
- **Seam** — where an interface lives; a place behaviour can be altered without editing in place. (Use this, not "boundary.")
- **Adapter** — a concrete thing satisfying an interface at a seam.
- **Leverage** — what callers get from depth.
- **Locality** — what maintainers get from depth: change, bugs, knowledge concentrated in one place.

Key principles (see [LANGUAGE.md](LANGUAGE.md) for the full list):

- **Deletion test**: imagine deleting the module. If complexity vanishes, it was a pass-through. If complexity reappears across N callers, it was earning its keep.
- **The interface is the test surface.**
- **One adapter = hypothetical seam. Two adapters = real seam.**

This skill is _informed_ by the project's domain model. The domain language gives names to good seams; ADRs record decisions the skill should not re-litigate.

## Process

### 1. Explore

Read the project's domain glossary and any ADRs in the area you're touching first.

For multi-repo or ecosystem-wide requests, start with a bounded inventory before spawning explorers: enumerate repos, read each repo's `AGENTS.md`/`CLAUDE.md`/`CONTEXT.md` when present, collect tracked-file/top-directory summaries with capped commands (prefer `git ls-files` over recursive filesystem walks), and split subagents into small repo batches with explicit read-only scope and output caps. If broad subagent batches time out, salvage the report from the bounded inventory and clearly mark partial evidence instead of rerunning the same unbounded scan.

Then use the Agent tool with `subagent_type=Explore` to walk the codebase. Don't follow rigid heuristics — explore organically and 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 styling, and **Mermaid via CDN** for diagrams where a graph/flow/sequence reliably communicates the structure. Mix Mermaid with hand-crafted CSS/SVG visuals — use Mermaid when relationships are graph-shaped (call graphs, dependencies, sequences), and hand-built divs/SVG when you want something more editorial (mass diagrams, cross-sections, collapse animations). Each candidate gets a **before/after visualisation**. Be visual.

For each candidate, the same template as before, but rendered as a card:

- **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 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 [LANGUAGE.md](LANGUAGE.md) vocabulary for the architecture.** If `CONTEXT.md` defines "Order," talk about "the Order intake module" — not "the FooBarHandler," and not "the Order service."

**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 in the card (e.g. a warning callout: _"contradicts ADR-0007 — but worth reopening because…"_). Don't list every theoretical refactor an ADR forbids.

See [HTML-REPORT.md](HTML-REPORT.md) for the full HTML scaffold, diagram patterns, and styling guidance.

Do NOT propose interfaces yet. After the file is written, ask the user: "Which of these would you like to explore?"

### 2A. If the user asks to turn candidates into issues

When the user asks for "all candidates", "all N", "one by one", or otherwise wants the report converted into GitHub work, create planning issues rather than implementation tasks. See `references/ecosystem-candidate-issue-expansion.md` for the restart-safe body shape and closeout pattern:

1. Verify the target GitHub repo/auth/branch before creating anything.
2. Search for likely duplicates using broad key nouns from the candidate set and inspect near matches; do not depend on a brittle exact-title query alone.
3. Create one issue per candidate, one at a time, with:
   - title scoped to the repo/domain and architecture action;
   - body containing summary, evidence basis, scope, out of scope, deliverables, acceptance criteria, plan-gate workflow, and related parent/report links;
   - `status:needs-plan` plus priority/category/domain labels that already exist in the repo taxonomy.
4. Keep the work explicitly planning-only. Do not start code changes from architecture-review issue creation.
5. Verify every created issue after creation: `OPEN` state, title, URL, labels, and `status:needs-plan`.
6. If there is a parent/portfolio issue, post one sequencing comment that lists recommended order and states that implementation remains blocked until each issue completes plan review and user approval.
7. Write a temp TSV or markdown index of created issue numbers/titles/URLs for restart-safe handoff; keep it outside repos unless the user asks for a tracked artifact.

### 2B. If the issue set becomes a staged plan wave

When architecture candidates expand into dependent GitHub issues, preserve the plan gate instead of treating patched drafts as approval. See `references/ecosystem-plan-wave-review-gates.md` for the detailed checklist.

- Use subagents for bounded patching, focused re-QA, and repo-batch inspection, but centrally verify every written plan/index change before advancing status.
- After workers patch upstream plans, run focused re-QA against the exact prior MAJOR findings and record the result in the plan artifact/index.
- Focused re-QA can clear substantive content blockers, but it does **not** replace formal provider-review artifacts.
- If required Claude/Codex/Gemini review artifacts, or explicit `UNAVAILABLE` artifacts, are missing, mark the plans `draft-provider-review-pending`, keep GitHub issues at `status:needs-plan`, and do not resume downstream planning.
- Never self-apply `status:plan-approved`; that transition remains user-only.

### 3. Grilling loop

Once the user picks a candidate, drop into a grilling conversation. Walk the design tree with them — constraints, dependencies, the shape of the deepened module, what sits behind the seam, what tests survive.

Side effects happen inline as decisions crystallize:

- **Naming a deepened module after a concept not in `CONTEXT.md`?** Add the term to `CONTEXT.md` — same discipline as `/grill-with-docs` (see [CONTEXT-FORMAT.md](../grill-with-docs/CONTEXT-FORMAT.md)). Create the file lazily if it doesn't exist.
- **Sharpening a fuzzy term during the conversation?** Update `CONTEXT.md` right there.
- **User rejects the candidate with a load-bearing reason?** Offer an ADR, framed as: _"Want me to record this as an ADR so future architecture reviews don't re-suggest it?"_ Only offer when the reason would actually be needed by a future explorer to avoid re-suggesting the same thing — skip ephemeral reasons ("not worth it right now") and self-evident ones. See [ADR-FORMAT.md](../grill-with-docs/ADR-FORMAT.md).
- **Want to explore alternative interfaces for the deepened module?** See [INTERFACE-DESIGN.md](INTERFACE-DESIGN.md).