Skip to main content
command CX Codex

isolated · serpro69, review-spec — Claude Code command: unbiased implementation review

isolated — a command that runs code review through a separate spec-reviewer sub-agent with no authorship bias. Surfaces gaps between documentation and implementation by comparing them from scratch.

Elastic-2.0 ★ 155 updated 2026-10-29

Install into Claude Code · Plugin marketplace All methods and platforms ↓
/plugin marketplace add serpro69/claude-toolbox
/plugin install kk@claude-toolbox
Open on GitHub

What isolated does

Available on Claude Code and Codex. Delegates analysis to a spec-reviewer sub-agent that checks implementation against feature docs in /docs/feat/wip/. Detects five problem types: spec deviations, extraneous code, missing implementation, outdated docs, and unclear requirements. Critical findings (spec violations, extra functionality) get detailed annotations with update suggestions; others are flagged briefly.

The command accepts a feature name or lists options from /docs/feat/wip/ if none is given. Output is organized by problem type and ready to feed back into development.

What it helps with

  • surface gaps between specification and code
  • eliminate author bias from the review process
  • classify issues by type and priority
  • update documentation based on findings

Example prompts for the agent

  • Review the jwt-auth feature against spec: /kk:review-spec:isolated jwt-auth
  • Run review for the only WIP feature in the project: /kk:review-spec:isolated
  • Check auth module implementation status: /kk:review-spec:isolated auth

How to install isolated

1Plugin marketplace

/plugin marketplace add serpro69/claude-toolbox
/plugin install kk@claude-toolbox

2Manually: put the files in place

.claude/commands/isolated.md

Call it with /isolated. For all projects use ~/.claude/commands/

Questions about isolated

How do I install isolated in Claude Code?

Run /plugin marketplace add serpro69/claude-toolbox, then /plugin install kk@claude-toolbox and place the .claude/commands/isolated.md file in your project.

What does isolated do?

It runs an independent code review using a sub-agent that compares your implementation against feature documentation. It identifies spec violations, extraneous code, and unclear requirements without the author's perspective.

Where should feature documentation be stored?

In /docs/feat/wip/{feature-name}/. The command automatically locates files there or prompts you to choose from existing features if you don't provide a name.

Similar slash commands

All slash commands for Code review and refactoring →

isolated by other authors comes first.

isolated serpro69 The command checks an answer by assigning separate verification questions to isolated sub-agents. Use it to review factual claims or investigate code in a project. CX Codex +1 ★ 155repo stars isolated serpro69 Runs an isolated multi-model code review on git changes. Evaluates security, compatibility, and design rules across separate reviewers. CX Codex +1 ★ 155repo stars gjalla-spec-review gjalla gjalla-spec-review — a skill that reads a plan or spec through several expert lenses and lists blocking issues before implementation starts. CX Codex +1 902installs local-codex daymade local-codex is a Claude Code skill that runs the local OpenAI Codex CLI as a non-interactive coding sub-agent via ChatGPT OAuth instead of API keys. CX Codex +1 412installs omc-review Yeachan-Heo omc-review is a Claude Code skill that checks finished code for defects, risks, and bloat before release. It ranks issues by severity without making fixes. CX Codex +1 309installs do-and-judge NeoLabHQ do-and-judge — a skill for Claude Code, Cursor, Codex and other AI coders: dispatches a sub-agent for the task, verifies the result through an LLM judge, and retries up to three times if the score falls below the threshold. CX Codex +1 1,300installs multi-agent-review riekelt multi-agent-review dispatches six automated reviewers across two model tiers to audit specs and plans before execution. Flags blockers, gaps, and risks prior to code generation. CX Codex +1 26,900installs prd-implementation-precheck zhaono1 The prd-implementation-precheck skill runs a PRD or feature spec through a checklist before implementation, surfacing edge-case gaps and blockers, then gets on with the build. CX Codex +1 718installs

From the same set30

The whole set →
default The default command checks an answer with Chain-of-Verification: it creates fact-checking questions, answers them separately, then revises the result. CX Codex +1 ★ 155repo stars default default validates code against feature design files to catch spec deviations, missing requirements, and outdated docs. CX Codex +1 ★ 155repo stars review-tool The review-tool skill checks tool configuration files against project schemas and reports violations. Use it when you need a full scan that continues after errors appear. CX Codex +1 ★ 155repo stars merge-docs merge-docs combines two doc sets for one feature into one: design, plan and tasks. Handy when design ran twice and you need a single version to build from. CX Codex +1 ★ 155repo stars isolated The command checks an answer by assigning separate verification questions to isolated sub-agents. Use it to review factual claims or investigate code in a project. CX Codex +1 ★ 155repo stars diff-skill Diffs two revisions of a skill's markdown instructions and flags the edits that weaken a rule or add avoidable complexity. Useful before you accept changes to a skill you rely on. CX Codex +1 ★ 155repo stars isolated Runs an isolated multi-model code review on git changes. Evaluates security, compatibility, and design rules across separate reviewers. CX Codex +1 ★ 155repo stars dependency-handling The dependency-handling skill makes an agent verify dependency details before writing code that uses them. It helps when adding or upgrading a library, API, IaC version, CRD, or container image. CX Codex +1 ★ 155repo stars

Project sponsors

Companies that keep this analytics open