agentsclimarketplace

Review as dev

Skill cloudfieldcz/cf-powers/skills/review-as-dev

Use when reviewing a technical analysis, design document, or specification from a developer perspective - checking technical feasibility, architecture, integration points, and performance/securityFrom its SKILL.md

Install
npx -y skills add cloudfieldcz/cf-powers --skill review-as-dev

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 7 stars7 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.

SKILL.md

3.8 KB, as published. Nobody here has run it

Review as Developer

Overview

Review a technical analysis or specification document from a senior developer perspective. Focus on whether the proposed solution is technically sound, feasible, and addresses architecture, performance, and security concerns.

Announce at start: "I'm reviewing this document as a Developer."

Input

  • Design + analysis document (typically docs/plans/YYYY-MM-DD-<topic>.md)
  • The actual codebase (read relevant files to verify claims in the analysis)

Review Checklist

1. Code Verification (DO THIS FIRST)

CRITICAL: Do not trust the analysis at face value. Read the actual codebase to verify:

  • Are the file:line references accurate?
  • Do the claimed existing patterns actually exist in the code?
  • Are the "files WITHOUT changes" truly unaffected?
  • Is the proposed approach compatible with the actual code?

2. Technical Feasibility

  • Is the proposed architecture realistic?
  • Are the technology choices appropriate?
  • Are there simpler alternatives that were not considered?
  • Do the estimated phases seem realistic?

3. Architecture Concerns

  • Does the solution fit the existing architecture?
  • Are SOLID principles respected?
  • Is the separation of concerns appropriate?
  • Are there coupling issues or circular dependencies?
  • Is the DI registration plan correct?

4. Missing Technical Details

  • Are database indexes defined for query patterns?
  • Are transaction boundaries specified?
  • Is concurrency handling addressed?
  • Are connection/resource lifetimes correct?
  • Are migrations reversible?

5. Integration Points

  • Are all integration points identified?
  • Are API contracts (method signatures) complete?
  • Will existing callers break?
  • Are backward compatibility concerns addressed?
  • Is the "files WITHOUT changes" section accurate?

6. Performance and Security

  • Are there N+1 query risks?
  • Is caching strategy appropriate?
  • Are there potential memory issues (large data sets)?
  • Is input validation sufficient?
  • Are authorization checks in place?
  • Are sensitive data handled correctly?

Output Format

Structure your review exactly like this:

# Dev Review: <topic>

## Summary
[1-2 sentence overall technical assessment]

## Architecture
- ✅ [Sound architectural decision]
- ❌ [Architecture concern -- explain why and suggest alternative]

## Technical gaps
- ❌ [Missing technical detail -- what needs to be specified]
- ❌ [Incorrect assumption -- what the code actually does at file:line]

## Integration risks
- ❌ [Integration point that may break]
- ❌ [Missing API contract detail]

## Performance / Security
- ❌ [Performance concern with suggested mitigation]
- ❌ [Security concern with suggested fix]

## Code verification
- ✅ [Verified: file:line reference is accurate]
- ❌ [Incorrect: file:line actually does X, not Y as claimed]

## Technical questions
1. [Question about implementation approach]
2. [Clarification about technical decision]

## Recommendations
1. [Actionable technical recommendation]
2. [Actionable technical recommendation]

## Verdict
**Technical quality:** [OK / Minor issues / Major issues]
**Recommended next step:** [Proceed to planning / Incorporate feedback / Needs architectural rework]

Key Principles

  • Verify against code -- Always read the actual files referenced in the analysis
  • Specific references -- Use file:line when pointing out issues
  • Pragmatic -- Focus on real risks, not theoretical perfection
  • Alternative proposals -- When flagging an issue, suggest a better approach
  • Security mindset -- Think about what could go wrong in production

Keep looking

Skills are one crate of 627,801. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.