Install
$ agentstack add skill-thixpin-claude-config-architecture-review ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Architecture Review
Judge structure by how safely the system can change, not by resemblance to a textbook pattern. Recommend the smallest set of moves that improves changeability; never propose a rewrite when an incremental path exists.
Scope
Reviews the system — modules, layers, and dependencies across the codebase.
Use for structural assessment: technical debt, module boundaries, coupling, duplication across modules, planning a refactor.
Do not use for:
- Reviewing a single diff or file for quality — use
code-quality-review. The same concerns (DRY, single responsibility) appear in both; the difference is altitude. Within one change, they are that skill's; across modules, they are this one's. - Implementing the refactor. This skill produces recommendations; carrying them out is ordinary work under
CLAUDE.md.
Method
- Map before judging. Identify the modules/layers, what depends on what, and where the domain logic actually lives. Use the dependency direction the code has, not the one the docs claim.
- Find where change hurts. Ask of recent or likely changes: how many places must be touched? Files that always change together but live far apart, or single files that change for unrelated reasons, mark the real problems.
- Diagnose against the checklist below.
- Recommend minimally. Each recommendation: the problem, the concrete evidence (files/modules), the smallest fix, and what it unblocks. Order by pain relieved per effort. Explicitly list what should be left alone.
Checklist
Duplication
- Same business rule or knowledge implemented in multiple places (the dangerous kind — they drift apart).
- Copy-paste variants of one workflow differing only in details that should be parameters.
- Do not flag incidental similarity between things that change for different reasons; merging those creates coupling, not reuse.
Single responsibility
- Modules or classes that change for unrelated reasons (e.g., a "service" doing validation, persistence, formatting, and notifications).
- God objects most of the codebase imports; files disproportionately large for their layer.
Layering & dependencies
- Lower layers importing upward (domain logic importing web/UI types; persistence importing controllers).
- Business rules trapped inside framework artifacts (controllers, handlers, views) where they can't be tested or reused.
- Circular dependencies between modules.
- Leaky boundaries: raw database rows, ORM entities, or wire DTOs flowing through layers that shouldn't know them.
Coupling & cohesion
- Feature logic scattered across layers such that one feature change touches every layer folder ("shotgun surgery").
- Shared mutable state or implicit ordering between modules.
Constraints
- Respect the project's existing architectural style; recommend making it consistent before recommending replacing it.
- Every proposed step must leave the system working — no recommendation may require a big-bang migration.
- If the architecture is sound, say so; do not manufacture findings.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: thixpin
- Source: thixpin/claude-config
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.