Skip to content
Back to skills

Ken Review

ASecurity

Review a diff for method violations: rot to rewrite, translate-only layers, unvouched deps, fancy over brute force. One line per finding.

  • 10 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added September 19, 2026
ai-agentsrustgogitsecurityperformance

Security analysis

A100/100

Scanned September 19, 2026

npx -y skills add rajnandan1/ken --skill ken-review --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Ken Review?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Ken Review
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/rajnandan1-ken-review/badge)](https://www.skillsdirectory.com/skills/rajnandan1-ken-review)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: ken-review
description: "Review a diff for method violations: rot to rewrite, translate-only layers, unvouched deps, fancy over brute force. One line per finding."
homepage: https://github.com/rajnandan1/ken
license: MIT
---

Review diffs for Thompson-mode violations. One line per finding: location,
what's wrong, the move. The diff's best outcome is fewer layers and a
smaller trusted base.

## Format

`L<line>: <tag> <what>. <the move>.`, or `<file>:L<line>: ...` for
multi-file diffs.

Tags:

- `rot:` unit on repeated patches; the next fix is a rewrite, not patch N+1. Name the rewrite's shape.
- `layer:` wrapper, adapter, or manager that adds no decision of its own. Delete it, move callers down a level.
- `unvouched:` dependency or pasted code nobody present can explain. Vouch for it (read it) or replace with a few lines you own.
- `fancy:` clever algorithm where the plain loop wins at this scale. Name the brute-force form; demand the measurement that justifies fancy.
- `ceremony:` process, abstraction, or config that serves the process itself. Remove it.

## Examples

❌ "This caching layer might benefit from reconsidering its abstraction
boundaries and perhaps simplifying the invalidation strategy."

✅ `L12-60: rot: third patch on this cache's invalidation. Rewrite on a plain dict + mtime check, ~25 lines.`

✅ `L4: unvouched: left-pad-like microdep imported unread. 3 lines inline, trusted base shrinks by one.`

✅ `svc.py:L88: layer: OrderManager delegates every call to OrderRepo. Delete, callers hit OrderRepo.`

✅ `L30-52: fancy: hand-rolled B-tree for 40 entries. Flat array + linear scan; ken: revisit when n > 10k measured.`

✅ `L71: ceremony: config flag nobody sets gates one constant. Inline the constant.`

## Scoring

End with the metrics that matter: `net: -<N> lines, -<M> layers, trusted base -<K> deps possible.`

If there is nothing to flag, say `Sound. Ship it.` and stop.

## Boundaries

Scope: Thompson-mode method violations only. Correctness bugs, security
holes, and performance are out of scope. Route them to a normal
review pass, not this one. A single smoke test or `assert`-based self-check
is the ken minimum, not ceremony, never flag it for removal.
Does not apply the fixes, only lists them.
"stop ken-review" or "normal mode": revert to verbose review style.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.