John Carmack-level implementation review via RepoPrompt or Codex. Use when reviewing code changes, PRs, or implementations. Triggers on /flow-next:impl-review.
Installation
Details
Usage
After installing, this skill will be available to your AI coding assistant.
Verify installation:
skills listSkill Instructions
name: flow-next-impl-review description: John Carmack-level implementation review via rp-cli for current branch changes. Use when reviewing code changes, PRs, or implementations. Triggers on /flow-next:impl-review.
Implementation Review Mode (CLI)
Conduct a John Carmack-level review of implementation changes on the current branch using RepoPrompt's context builder and chat.
Role: Code Review Coordinator (NOT the reviewer) Tool: rp-cli for context building and chat delegation
Input
Arguments: #$ARGUMENTS
Format: [additional context, focus areas, or special instructions]
Example: /flow-next:impl-review focus on the auth changes, ignore styling
Reviews all changes on the current branch vs main/master.
Setup
FLOWCTL="${CLAUDE_PLUGIN_ROOT}/scripts/flowctl"
FIRST: Determine Review Mode
Check: which rp-cli >/dev/null 2>&1
If NOT available: inform user rp-cli is required for this skill.
Option Parsing (skip questions if found in arguments)
Parse the arguments for these patterns. If found, use them and skip the question:
Review mode:
--mode=rpor--rpor "rp chat" or "repoprompt" → RepoPrompt chat--mode=exportor--exportor "export" or "external llm" → export for external LLM
If options NOT found
If review mode was already chosen earlier in this conversation (e.g., user answered "2a" or "2b" during /flow-next:plan or /flow-next:work setup):
→ Use that mode, don't ask again.
If invoked directly without prior context, output this text (do NOT use AskUserQuestion tool):
Both modes use RepoPrompt for context building (builder, file selection, codemaps).
The difference is where the review happens:
Review mode:
a) RepoPrompt chat (default) — review via rp-cli chat
b) Export for external LLM — export context file for ChatGPT, Claude web, etc.
(Reply: "a", "b", "export", or just tell me)
Wait for user response. Parse naturally.
Context Sources
The workflow gathers context from:
- Git diff and commit messages
- Epic/task specs if a Flow ID is known
- PRD/architecture docs
If you know which Flow epic/task this work relates to, include context:
$FLOWCTL show <id> --json
$FLOWCTL cat <id>
Critical Requirement
DO NOT REVIEW CODE YOURSELF – you are a coordinator, not the reviewer.
Your job is to:
- Use
rp-cli -e 'windows'to find the RepoPrompt window - Use
rp-cli -w <id> -e 'builder ...'to build context - Use
rp-cli -w <id> -e 'chat ...'to execute the review
The RepoPrompt chat conducts the actual review with full file context.
Workflow
Read workflow.md and follow each phase in order. Phases include change identification, context building, and review execution.
rp-cli Reference
Read rp-cli-reference.md for command syntax and examples.
More by gmickel
View allCarmack-level plan review via rp-cli context builder + chat. Use when reviewing implementation plans, architecture docs, or design specs. Triggers on mentions of plan review, design review, or architecture review.
Manage git worktrees (create/list/switch/cleanup) and copy .env files. Use for parallel feature work, isolated review, clean workspace, or when user mentions worktrees.
Carmack-level plan review via RepoPrompt or Codex. Use when reviewing Flow epic specs or design docs. Triggers on /flow-next:plan-review.
Execute a plan file or Beads epic systematically with git setup, task tracking, quality checks, and commit workflow. Use when implementing a plan, working through a spec, following documented steps, or executing a Beads issue ID (e.g., bd-123, gno-45, app-12).
