By anortham
Miller-powered development workflow with subagent-driven execution, inline review, and Miller-first codebase orientation.
Use when planning or reviewing non-trivial code changes, refactoring architecture, evaluating module boundaries, repeated findings reveal coupling, tests are hard to write because interfaces are unclear, or the user asks for codebase design improvements, complexity reduction, deduplication, or cleanup of existing code.
You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.
Call the Claude Code CLI (`claude -p`) as a fresh Claude instance for second opinions, code review, and adversarial review. Use this skill whenever the user says "ask claude", "fresh claude review", "second opinion from another claude", "have another claude look at this", "delegate to a fresh claude", or any variation suggesting they want a second Claude instance's perspective. The value is a fresh session and independent prompt framing, not a different model.
Call OpenAI's Codex CLI for second opinions, code review, adversarial review, and task delegation. Use this skill whenever the user says "ask codex", "get codex's take", "codex review", "second opinion", "have codex look at this", "delegate to codex", "deep review", "adversarial review", or any variation suggesting they want Codex's perspective. Also trigger when the user wants a fresh take from a different model on code, architecture, or implementation decisions.
Use when the user wants two models to alternate finding, verifying, and fixing problems until none remain above an agreed bar — "have codex verify my findings and find more", "go back and forth until clean", "loop until you both agree there are no problems left", or when an audit should run to convergence under /loop or another goal-driven runner. Also use for a bounded pre-implementation Doubt Pass when a design decision needs adversarial challenge before building — architecture risk rated medium/high, or the user says "doubt this" / "challenge this decision".
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Miller-powered development workflow for Claude Code, Cursor, Codex, OpenCode, Copilot CLI, and Gemini CLI.
Razorback is a skill set for every major coding-agent harness, diverged from Superpowers to add Miller MCP for token-efficient codebase orientation. Plan execution runs through subagent-driven-development on harnesses that support delegation, and executing-plans where delegation is unavailable.
AI-assisted development burns tokens on repetitive codebase exploration. Every agent and subagent re-discovers the same code through Glob/Grep/Read chains. Razorback solves this two ways:
search, context, inspect, trace, impact, and workspace — that return targeted context in 1-2 calls instead of 5-8.multi_agent = true in ~/.codex/config.toml so parallel execution skills can dispatch subagentsRAZORBACK.md for project-specific razorback policy, such as model routing and verification tiersRazorback is a pure-content plugin (skills, commands, hooks) — no build step or runtime dependencies required. Install paths below are all "pull from GitHub"; most harnesses support one-liner installs.
/plugin marketplace add anortham/razorback
/plugin install razorback@razorback
Scope to a specific project instead of user-wide:
/plugin install razorback@razorback --scope project
Or clone and load locally for development:
git clone https://github.com/anortham/razorback.git
claude --plugin-dir /path/to/razorback
Cursor has a plugin marketplace, but razorback isn't listed there yet. For now, clone and point Cursor at the local checkout via its plugin-load mechanism:
git clone https://github.com/anortham/razorback.git ~/path/to/razorback
# Then in Cursor: add ~/path/to/razorback to your plugin paths
The plugin manifest at .cursor-plugin/plugin.json declares skills, agents, commands, and hooks. Cursor marketplace submission is tracked as a follow-up — it will just be a one-liner when live.
Tell Codex:
Fetch and follow instructions from https://raw.githubusercontent.com/anortham/razorback/refs/heads/main/.codex/INSTALL.md
Codex uses native skill discovery, so installation is a clone plus a symlink at ~/.agents/skills/razorback. Delegated execution skills (subagent-driven-development, dispatching-parallel-agents) require Codex's multi_agent feature.
Detailed docs: .codex/INSTALL.md
Tell OpenCode:
Fetch and follow instructions from https://raw.githubusercontent.com/anortham/razorback/refs/heads/main/.opencode/INSTALL.md
Detailed docs: .opencode/INSTALL.md
copilot plugin marketplace add anortham/razorback
copilot plugin install razorback@razorback
Copilot CLI reads the same .claude-plugin/marketplace.json that Claude Code uses. Named plugin agents (like razorback:code-reviewer) are auto-discovered.
gemini extensions install https://github.com/anortham/razorback
Gemini loads gemini-extension.json + GEMINI.md at session start, which pulls in the using-razorback skill and the Gemini tool mapping. Subagent dispatch goes through Gemini's invoke_agent tool with the built-in generalist agent (parallel by default), so subagent-driven-development and dispatching-parallel-agents work the same way they do on the other harnesses.
Once loaded, razorback works automatically. The bootstrap path varies by harness:
npx claudepluginhub anortham/razorback --plugin razorbackCode intelligence server: search, navigation, and refactoring across 34 languages
Cross-client MCP memory with checkpoints, recall, briefs, and standups for AI-assisted development
Agent memory persistence - remember and recall across sessions
Composite workflows that orchestrate Julie (code intelligence), Sherpa (workflow guidance), and Goldfish (persistent memory) for powerful development patterns. Includes TDD Powerhouse, Bug Detective, Smart Session Start, and more.
Local SQLite-backed code intelligence, content search, and workspace freshness for coding agents.
Complete collection of battle-tested Claude Code configs from an Anthropic hackathon winner - agents, skills, hooks, and rules evolved over 10+ months of intensive daily use
oh-my-zsh for Claude Code — 11 agents, 33 commands, 24 skills, 15 hooks + 9 examples (21 events), 9 rules, 4 MCP (minimal: playwright, context7, jina-reader, [email protected]). One-line install: curl -fsSL https://raw.githubusercontent.com/sangrokjung/claude-forge/main/install.sh | bash
One plugin to bridge and delegate across Claude Code, Codex CLI, and Gemini CLI — single-source AGENTS.md, shared skills, mirrored hooks and MCP servers, and full Claude↔Codex bidirectional delegation.
70+ Claude Code slash commands across 12 development phases with Dagger-based safety system, multi-dimensional validation, and specialized agents
Task-focused agents for test, review, debug, docs, CI, security, refactoring, research, performance, and search-replace — with teammate and subagent role guidance
Multi-agent /workflow development pipeline (planner → plan-review → coder → code-review) with typed handoff contracts, lifecycle hooks, and MCP servers.