SuperAPI AI Directory

thermo-nuclear-code-quality-review

Testing & quality

View repository

Updated 10-15 21:09

About this skill

agentic-workflows-blueprint.workflow.thermo-nuclear-code-quality-review

Goal

Perform an unusually strict review focused on implementation quality, maintainability, abstraction quality, and codebase health.

Scope

  • Applies to: Any code change or pull request diff across the codebase.
  • Does not cover: Direct code modification or executing fixes (use thermo-fix).

Triggers

  • "Run thermo-nuclear review"
  • "/thermo-nuclear-code-quality-review"
  • "Deep code quality audit"
  • "Harsh maintainability review"

Inputs

  • diffScope: Files or commit range to review.
  • existingRootDoc: Main instruction file (AGENTS.md, CLAUDE.md, etc.).

Invariants

  1. Be ambitious about structural simplification: Search for "code judo" moves that make implementation simpler, smaller, and cleaner.
  2. 1000 line limit: Do not let a PR push a file past 1k lines without strong justification.
  3. No random spaghetti growth: Reject ad-hoc conditionals or one-off branches inserted into unrelated flows.
  4. Clean design over working code: Do not approve merely because tests pass if the structure is messier.
  5. No silent fallbacks: Ensure boundary cleanliness and explicit error/state contracts.

Procedure

  1. Identify all changed files in diffScope.
  2. Evaluate against the Non-Negotiable Additional Standards:
    • Structural simplification & Code Judo opportunities.
    • Decomposition of files approaching 1k lines.
    • Elimination of ad-hoc conditionals or leaky abstractions.
    • Type contract and boundary cleanliness.
  3. Produce a Severity-Ranked Finding Table:
    • Blocker: Structural regressions, 1k+ line file growth, severe spaghetti.
    • High: Major maintainability or boundary issues.
    • Medium: Moderate cleanup or abstraction improvements.
    • Low: Informational nits.
  4. Output specific, actionable remedies for every Blocker and High finding.

Outputs

  • Severity-ranked finding table appended to walkthrough.md or conversation output.
  • Final verdict: APPROVED or FINDINGS REMAIN.

Review gate

  • No unaddressed Blocker findings.
  • No unaddressed High findings.
  • Implementation preserves architecture contracts.

References

  • ../../SKILL.md
  • ../thermo-fix/SKILL.md
  • ../radioactive/SKILL.md