improve-codebase-architecture
Scan a codebase for deepening opportunities, present them as a visual HTML report, then grill through whichever one you pick.
- License 许可
- MIT
- 最近更新
- 2026-09-18
- Stars
- ★ 1
- Forks
- 1
01 前言 Front Matter
- name
- improve-codebase-architecture
- description
- Scan a codebase for deepening opportunities, present them as a visual HTML report, then grill through whichever one you pick.
- disable-model-invocation
- true
02 资源文件 1
- 参考文档HTML-REPORT.md6.6 KB
03 SKILL.md 正文
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.
This command is informed by the repo's existing decisions and built on a shared design vocabulary:
- Run the
/codebase-designskill for the architecture vocabulary (module, interface, depth, seam, adapter, leverage, locality) and its principles (the deletion test, "the interface is the test surface", "one adapter = hypothetical seam, two = real"). Use these terms exactly in every suggestion — don't drift into "component," "service," "API," or "boundary." - The repo's own domain vocabulary (its
GLOSSARY.mdif present,CLAUDE.md, subsystem READMEs, and any in-flight_specs/bundle for the area) gives names to good seams. The main-bundle decision register (_specs/<scope>/main/spec.md) records architecture decisions this command must not re-litigate — it is the "what is still true" surface, read first.
Process
1. Explore
Read the repo's domain vocabulary first — its GLOSSARY.md (if present), CLAUDE.md, subsystem READMEs, and any active _specs/<repo>/<DATE>-<handle>/ bundle for the area you're touching. Then read the main-bundle decision register (_specs/<scope>/main/spec.md) so you don't re-suggest a decision already recorded there (and check the active bundle's spec.md if the area is mid-flight).
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, 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 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 the repo's own vocabulary for the domain (drawn from GLOSSARY.md if present, CLAUDE.md, subsystem READMEs, and _specs/), and the /codebase-design vocabulary for the architecture. If the repo calls it "Order intake," talk about "the Order intake module" — not "the FooBarHandler," and not "the Order service."
Decision-register conflicts: if a candidate contradicts a recorded decision, only surface it when the friction is real enough to warrant reopening that decision. Mark it clearly in the card (e.g. a warning callout: "contradicts §<id> 'single intake seam' in the decision register — but worth reopening because…"), citing the entry by its §<id> header. Don't list every theoretical refactor a past decision forbids.
See 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?"
3. Grilling loop
Once the user picks a candidate, run the /grilling skill to 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 the repo hasn't named yet? Record the new term in
GLOSSARY.md(the canonical domain-vocabulary surface) so future sessions name it consistently — lazily creatingGLOSSARY.mdwith a one-lineCLAUDE.mdpointer if it doesn't exist yet, per the/domain-modelingconvention. Otherwise just use it consistently for the rest of this session. - User rejects the candidate with a load-bearing reason? Offer to record it as a load-bearing decision so future architecture reviews don't re-suggest it, per DECISION-FORMAT.md: a durable, repo-level architecture decision lands in the main-bundle decision register (
_specs/<scope>/main/spec.md— the ADR-equivalent surface); a decision scoped to an in-flight planning bundle lands in that bundle'sspec.md. Frame the offer as: "Want me to record this as a load-bearing decision 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. - Want to explore alternative interfaces for the deepened module? Run the
/codebase-designskill and use its design-it-twice parallel sub-agent pattern.
04 同仓库更多 Skills
prompt-optimization
Use when the user asks to optimize, improve, harden, or tighten a prompt; when /optimize-prompt is invoked; when a user hands over a rough prompt to be made AI-ready; when a prompt is described as vague, incomplete, or missing guardrails; or when the user asks for help making a prompt "production-ready" or "model-ready."
pr-review-loop
Use when driving a multi-pass PR review to a merge bar; when the user invokes /pr-review-loop; when a review pass returned critical findings that need fixing followed by another review; when iterative PR review-fix cycles need to run mostly hands-off.
handoff
Compact the current conversation into a handoff document for another agent to pick up.
grilling
Interview the user relentlessly about a plan or design. Use when the user wants to stress-test a plan before building, or uses any 'grill' trigger phrases.
grill-with-docs
A relentless interview to sharpen a plan or design, which also maintains the repo's GLOSSARY.md and logs load-bearing decisions as you go.
grill-me
A relentless interview to sharpen a plan or design.