Architecture review
Skill nledford/engineering-review-board/skills/architecture-review
A collection of AI agent skills I have written
npx -y skills add nledford/engineering-review-board --skill architecture-reviewAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Review architectural boundaries, dependency direction, ports and adapters, Clean/Onion/Hexagonal/DDD hybrids, and modular monolith structure. Use for architecture-focused audits of existing systems or changes; do not use for ordinary implementation or architecture design without a review request.
SKILL.md
1.6 KB, as published. Nobody here has run it
Architecture Review Skill
Use this skill as an architecture-specific review lens for an existing system,
proposal, or diff. Always load
review-verification-protocol
before reporting findings. For repository change reviews, also load
code-review.
Do not use it for simple code changes with no boundary impact. Use the matching Clean, Hexagonal, Onion, or DDD skill when the primary task is designing or implementing an architecture rather than reviewing one.
Workflow
- Read project-local guidance, architecture docs, module structure, and the relevant diff before inferring the intended architecture.
- Map dependency direction, module and bounded-context boundaries, ports, adapters, and framework or persistence edges.
- Identify concrete leakage, cycles, misplaced policy, and boundary bypasses.
- Distinguish intentional hybrid choices from accidental drift; prefer coherent, testable structure over pattern purity.
- Verify findings against imports, call paths, tests, or build rules. Record skipped checks and residual architectural risk.
Output
Return findings ordered by severity with file-level evidence, impact, the smallest practical correction, skipped validation, and residual risk.