mayankmankhand

review-full

Pre-release cross-domain review with go/no-go recommendation. Use for release gates, major milestones, or when multiple domains changed and you need a single assessment.

review-browser

Browser QA review - drives a headless browser to test a running web app. Use for verifying visual layout, interactive flows, error states, and runtime behavior.

review-code

Code quality review - security, logic, performance, maintainability. Use for reviewing code changes like bug fixes, features, refactors, scripts.

review-commands

Slash command prompt review - prompt quality, workflow completeness, cross-command consistency. Use for reviewing .claude/commands/*.md or .claude/skills/*/SKILL.md files.

Copy clarity and reader orientation review - checks whether content orients a fresh reader. Use for reviewing web pages, blog posts, landing pages, guides, prototypes, and any reader-facing deliverable.

review-deps

Dependency and supply chain security review. Checks for known CVEs, outdated packages, single-maintainer risks, and license issues.

review-plan

Plan compliance review - checks if implementation matches the plan. Use for verifying feature completeness, scope drift, and quality gates against a PLAN-*.md file.

review-security

Application security review of a code change - the diff-catchable vulnerability classes (secrets, injection, XSS, path traversal, SSRF, unsafe deserialization, weak crypto) reviewed through an adversarial lens. Use when code changed and you want a security pass; runs automatically inside /review on any code change. For dependency CVEs use /review-deps; for a deep whole-repo audit use /security-audit.

review-ux

UX quality review - usability, accessibility, user flows, and how the UI feels. Use for evaluating user experience from code, markup, specs, and screenshots.

security-audit

Deep, on-demand whole-repo security audit - entry points, authorization on every route, crypto inventory, and a recommended secret scan across git history. Use deliberately for a feature milestone, a pre-release gate, or when /review nudges you after a change to auth, crypto, secrets, or a new external entry point. For a fast per-change pass use /review-security; for dependency CVEs use /review-deps.

error-analysis

Group the correction ledger's open codes into categories, count them, and rank what you actually keep correcting. The second half of the two-stage coding started by /tk:document's capture stage. Read-only against the ledger; never publishes and never sends anything off the machine.

playground

Generate throwaway, self-contained HTML files for in-the-loop decisions. Use when the user needs to see, drag, toggle, slide, or compare options to decide. Static-default; interactive only when it changes the decision.

project-context

Provides project context (tech stack, conventions, key rules) to subagents. Not a user command - loaded by the orchestrator when dispatching review subagents.

setup

Seed a project for the LLM Peer Review plugin, or migrate a copy-install onto it - the short rules file, gitignore lines, folders, LESSONS.md, DESIGN-PROFILE.md, the marketplace pointer and permissions; managed toolkit files are removed (backed up first) and every custom file is kept. Run once per project after installing the plugin.

upgrade

After /plugin update (or right after a /setup migration), audit this project's own commands, skills, agents, rules, CLAUDE.md, and the other files its sessions read against the toolkit conventions that changed between the last audited version and the installed one, plus C-7 and C-9 to C-12, which run on every upgrade, through the normal M2 audit and auto-fix loop. Opens the cycle's issue, fixes what drifted (the audit rerun decides FIXED, by each finding's key), runs one sample cycle, stamps the state file, and chains into /document.

learning-opportunity

Pause development to learn a concept at 3 levels of depth. Triggers when unfamiliar patterns, new frameworks, or complex concepts appear during development.

dispatch-contract

The output contract every finder agent preloads - JSONL findings or the literal NO FINDINGS, single pass, never audits its own findings. Not a command; it exists to be preloaded by the review finder agents.

review-browser-criteria

The browser qa review criteria, reading budget, severity anchors, finding ids, and finding contract, preloaded into the review-browser-finder agent. Not a command; the direct-run skill is /tk:review-browser.

review-code-criteria

The code review criteria, reading budget, severity anchors, finding ids, and finding contract, preloaded into the review-code-finder agent. Not a command; the direct-run skill is /tk:review-code.

review-commands-criteria

The slash command review criteria, reading budget, severity anchors, finding ids, and finding contract, preloaded into the review-commands-finder agent. Not a command; the direct-run skill is /tk:review-commands.

review-copy-criteria

The copy review criteria, reading budget, severity anchors, finding ids, and finding contract, preloaded into the review-copy-finder agent. Not a command; the direct-run skill is /tk:review-copy.

review-deps-criteria

The dependency review criteria, reading budget, severity anchors, finding ids, and finding contract, preloaded into the review-deps-finder agent. Not a command; the direct-run skill is /tk:review-deps.

review-plan-criteria

The plan compliance review criteria, reading budget, severity anchors, finding ids, and finding contract, preloaded into the review-plan-finder agent. Not a command; the direct-run skill is /tk:review-plan.

review-security-criteria

The security review criteria, reading budget, severity anchors, finding ids, and finding contract, preloaded into the review-security-finder agent. Not a command; the direct-run skill is /tk:review-security.

review-ux-criteria

The ux review criteria, reading budget, severity anchors, finding ids, and finding contract, preloaded into the review-ux-finder agent. Not a command; the direct-run skill is /tk:review-ux.