improve-codebase-architecture
Find architectural friction in a change set, evaluate module depth and test boundaries, and apply worthwhile scoped refactors in a pipeline architecture phase.
Skill by CraigThomasParsons
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/craigthomasparsons-improve-codebase-architecture
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: testing, data
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/improve-codebase-architecture
What's inside
- Skill name
improve-codebase-architecture- Description
- Find architectural friction in a change set, evaluate module depth and test boundaries, and apply worthwhile scoped refactors in a pipeline architecture phase. Also supports standalone architecture exploration.
- License
- MIT
- Instructions
- 14 lines in SKILL.md · view on GitHub
Files (3)
- SKILL.md
- openai.yaml · agents/
- HTML-REPORT.md · references/
SKILL.md (first lines)
# Improve codebase architecture Read CONTEXT.md and applicable ADRs before suggesting changes; preserve the domain vocabulary and existing decisions. Use this vocabulary: a **module** hides implementation behind an **interface**; **depth** is useful hidden complexity relative to interface complexity; a **seam** is an observable test boundary; an **adapter** connects external infrastructure; **locality** keeps related concepts together; **leverage** is the amount of useful behavior improved by a small interface change. Avoid extracting functions solely to expose internals to tests. One adapter often represents a hypothetical seam; multiple real consumers provide stronger evidence for abstraction. 1. Scope the scan to the issue's change set and closely affected modules. For standalone exploration, use the user's direction or recently changing areas. Identify shallow interfaces, knowledge spread across files, leaking boundaries, and behavior that is hard to test through the public interface. 2. Apply the deletion test: would removing the proposed abstraction concentrate complexity, or merely move it? Prefer designs that hide meaningful complexity and keep related behavior together. Respect ADRs; don't implement a conflicting design without an explicit decision. 3. For each candidate record files, observed problem, proposed change, public interface/test effects, and strength (Strong / Worth exploring / Speculative). Separate measured friction from hypothetical future needs. 4. In slice-pipeline ARCHITECTURE mode, apply the worthwhile candidates that fit the issue's scope. Keep behavior unchanged, validate it, commit and push each refactor. Record too-large candidates with a rationale and assigned follow-up issue. Do not stop at an assessment and do not enter REVIEW in this context. No candidate is a valid result when the current design needs no change; do not manufacture edits. 5. In standalone exploration mode, present candidate before/after diagrams when useful and ask which substantial direction to explore before implementing a new design. Use a temporary HTML report only when it helps the user compare candidates; [references/HTML-REPORT.md](references/HTML-REPORT.md) provides the original report patterns. Do not automatically open a browser from an unattended worker. Record newly agreed domain terms and lasting decisions in the repository's domain/ADR documents within the authorized scope. Pipeline packets already define the scope; no repeated design-selection interview is needed for ordinary behavior-preserving improvements. Decisions that change product behavior or approved architecture remain the operator's.
From CraigThomasParsons/TheGiteaCodeFactory, MIT.
More from CraigThomasParsons/TheGiteaCodeFactory
- 0.589score
supervise-agent-pipeline
CraigThomasParsons
Supervise a durable multi-phase agent worker, classify execution state, handle bounded approvals and provider fallback, and advance only after validating…
SkillAFreshness: A★ 1Claude Code +9 - 0.586score
tmux-pipeline
CraigThomasParsons
Execute one issue in isolated worktrees through implementation, simplification, architecture, review and PR phases.
SkillAFreshness: A★ 1Claude Code +9 - 0.581score
pr-review-resolve-loop
CraigThomasParsons
Review and repair bounded Gitea PR batches and merge only current-head cleared PRs under the project merge policy.
SkillAFreshness: A★ 1Claude Code +9 - 0.581score
dispatcher
CraigThomasParsons
Select a dependency-ready feature slice, audit BDD coverage, and prepare issue-backed work packets.
SkillAFreshness: A★ 1Claude Code +9 - 0.581score
slice-pipeline
CraigThomasParsons
Deliver an issue-backed feature slice through serial tmux phases with durable evidence and resumable supervision.
SkillAFreshness: A★ 1Claude Code +9 - 0.579score
simplify
CraigThomasParsons
Simplify an issue's changed code without changing behavior, then verify and push the scoped cleanup.
SkillAFreshness: A★ 1Claude Code +9 - 0.576score
scrum-master
CraigThomasParsons
Supervise issue-backed delivery, identify stalled work and blockers, and reconcile queue progress.
SkillAFreshness: A★ 1Claude Code +9 - 0.576score
coordinate-cross-project-work
CraigThomasParsons
Plan dependency-ordered repository work and reconcile cross-project evidence without editing product repositories.
SkillAFreshness: A★ 1Claude Code +9 - 0.576score
moonlighter
CraigThomasParsons
Coordinate factory work through Moonlighter intake and authoritative job claims.
SkillAFreshness: A★ 1Claude Code +9 - 0.576score
pr-resolve-loop
CraigThomasParsons
Repair current-revision Gitea PR findings and request independent re-review.
SkillAFreshness: A★ 1Claude Code +9 - 0.576score
pr-review-resolver
CraigThomasParsons
Address human feedback on an existing PR branch and draft replies without posting them.
SkillAFreshness: A★ 1Claude Code +9 - 0.576score
release-preparation
CraigThomasParsons
Prepare an evidence-backed GitHub mirror and draft release for a completed Gitea milestone.
SkillAFreshness: A★ 1Claude Code +9
Signals
- Stars
- 1
- Forks
- 0
- Open issues
- 3
- Last commit (this folder)
- Sep 28, 2026 (1 month ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from CraigThomasParsons
- 0.589score
supervise-agent-pipeline
CraigThomasParsons
Supervise a durable multi-phase agent worker, classify execution state, handle bounded approvals and provider fallback, and advance only after validating…
SkillAFreshness: A★ 1Claude Code +9 - 0.586score
tmux-pipeline
CraigThomasParsons
Execute one issue in isolated worktrees through implementation, simplification, architecture, review and PR phases.
SkillAFreshness: A★ 1Claude Code +9 - 0.581score
slice-pipeline
CraigThomasParsons
Deliver an issue-backed feature slice through serial tmux phases with durable evidence and resumable supervision.
SkillAFreshness: A★ 1Claude Code +9 - 0.581score
pr-review-resolve-loop
CraigThomasParsons
Review and repair bounded Gitea PR batches and merge only current-head cleared PRs under the project merge policy.
SkillAFreshness: A★ 1Claude Code +9 - 0.581score
dispatcher
CraigThomasParsons
Select a dependency-ready feature slice, audit BDD coverage, and prepare issue-backed work packets.
SkillAFreshness: A★ 1Claude Code +9 - 0.579score
simplify
CraigThomasParsons
Simplify an issue's changed code without changing behavior, then verify and push the scoped cleanup.
SkillAFreshness: A★ 1Claude Code +9
Cite this page
improve-codebase-architecture skill. Agent Mart. https://agentmart.dev/skills/craigthomasparsons-improve-codebase-architecture (last updated 2026-10-24)