Code review

Skill shinzoxD/knackbox/skills/coding/code-review

Curated, auditable, benchmark-ready Agent Skills library for Claude Code, Codex, OpenCode, Cursor, and more.

Install

npx -y skills add shinzoxD/knackbox --skill code-review

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

2 things to look at

28 days oldThe repository was created 28 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
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 code changes for bugs, security issues, and maintainability with severity-tagged, actionable findings. Use whenever the user shares a diff, pull request, or code snippet and asks for review, feedback, "does this look right", "find issues", or "check my code" — even for small snippets or partial files.

The file declares its own license as Apache-2.0. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

2.8 KB, 563 tokens by cl100k_base, as published. Nobody here has run it

Code Review

Review code the way a senior engineer would: find the problems that matter, explain why they matter, and propose a concrete fix for each. The goal is a better change merged faster — not a list of nitpicks.

Review order

Work through these in priority order and tag every finding:

[blocking] — bugs, security vulnerabilities, data loss, race conditions, broken error handling. Must be fixed before merge.
[important] — performance traps, missing edge cases, misleading names, absent tests for risky logic. Should be fixed, can be discussed.
[nit] — style and taste. Optional; prefix so the author can skip.

Rules

Every finding names the location (file/function/line), explains the consequence, and shows a fix — a code suggestion, not just "handle this better".
Review the change, not the codebase. Pre-existing problems outside the diff get at most one grouped note at the end.
Don't raise style issues a formatter or linter would catch; assume tooling exists unless the user says otherwise.
If the code is good, say so specifically ("the retry logic correctly caps backoff") — one line, no padding.
Never invent behavior of functions you can't see. Say "if save() can raise, this leaks the lock" rather than asserting it does.
End with a verdict: approve, approve with nits, or request changes, plus a one-sentence reason.

Output format

## Review: <one-line summary of the change>

**Verdict:** request changes — <reason>

### Findings
1. [blocking] `auth.py:42` — Token compared with `==`, enabling timing
   attacks. Use `hmac.compare_digest(token, expected)`.
2. [important] `auth.py:58` — No test covers the expired-token path.
3. [nit] `utils.py:12` — `d` → `retry_delay_seconds`.

### Out of scope (pre-existing)
- Module has no type hints; consider a follow-up.

Edge cases

No context on language/framework versions: state your assumption in one line and review anyway.
Huge diff (>500 lines): review architecture and the riskiest files first, then offer to go file-by-file.
User asks "is this secure?": review against OWASP-style basics (injection, authz, secrets, input validation) and say explicitly which you checked.

Gives 0 of the 12 instructions most code review skills give in 563 tokens

Counted across 610 of the 674 authors here whose files we hold, read 2026-08-06

push back with technical reasoning if wrongin 60 of 610, across 24 files
ask for clarification on unclear itemsin 51 of 610, across 16 files
fix critical issues immediatelyin 45 of 610, across 29 files
implement one item at a timein 45 of 610, across 11 files
group findings by severityin 44 of 610, across 43 files
verify feedback against the codebasein 42 of 610, across 8 files
dispatch a code reviewer subagentin 39 of 610, across 23 files
fix important issues before proceedingin 37 of 610, across 22 files
test each fix individuallyin 35 of 610, across 7 files
reply in github comment threadsin 33 of 610, across 5 files
check for security vulnerabilitiesin 31 of 610, across 27 files
factualize corrections without over-explainingin 30 of 610, across 2 files

Said here and by no other author read

state the consequence of each finding
name the location of each finding
group pre-existing problems at the end
do not raise linter or formatter issues

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.