By STRML
Run parallel AI reviews from multiple models (Codex, Gemini, Kimi, Qwen, etc.) to analyze plans, debate contradictions, and synthesize a consensus verdict—with personality-driven reviewers like Skeptic and Architect.
Check acpx CLI installation, validate debate-acpx.json config, probe each configured agent, and print permission allowlist for unattended operation.
Run ALL configured AI reviewers in parallel via acpx, synthesize feedback, debate contradictions, and produce a consensus verdict. Configure reviewers in ~/.claude/debate-acpx.json.
Interactive Claude review — choose personalities and model before review starts. Shortcut for claude-review in interactive mode.
Shortcut for claude-review with the Skeptic pair (Fable + Opus) plus the Architect. Accepts --model sonnet to override the Architect's default opus.
Run Claude reviewer(s) on the current plan. Defaults to a Skeptic pair (Fable + Opus, model-tuned prompts). Use claude-double-review to add the Architect, claude-custom-review for interactive picker.
Runs pre-commands
Contains inline bash commands via ! syntax
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.
Get a second (and third, and fourth) opinion on your implementation plan before you write a line of code. debate sends your plan to multiple AI reviewers in parallel, synthesizes their feedback, has them argue out contradictions, and produces a consensus verdict.
v2 is a ground-up rewrite. Everything now runs through acpx — a single unified CLI that talks to any coding agent. No more managing individual CLIs, session files, or API keys per provider. One config, any combination of models.
# 1. Install the plugin
/plugin marketplace add STRML/cc-debate
/plugin install debate@cc-debate
# 2. Install acpx
npm install -g acpx@latest
# 3. Configure your review panel (interactive)
/debate:acpx-setup
# 4. Run a review
/debate:all
Restart Claude Code after installing the plugin.
You: /debate:all
Claude: Running parallel review via acpx...
codex → agent: codex (120s)
gemini → agent: gemini (240s)
mercury → agent: mercury (120s)
## Codex Review — Round 1
The retry logic in Step 4 doesn't handle the case where...
VERDICT: REVISE
## Gemini Review — Round 1
Missing error handling when the API is unavailable...
VERDICT: REVISE
## Mercury Review — Round 1
Unstated assumption: this plan assumes the temp directory is writable...
VERDICT: REVISE
## Synthesis
Unanimous: all reviewers flagged missing error handling
Unique to Codex: retry logic gap in Step 4
Unique to Mercury: temp directory writability assumption
## Final Report
VERDICT: REVISE — 3 issues to address
Claude: Revising plan...
Claude: Re-submitting to all reviewers...
## Codex Review — Round 2 → VERDICT: APPROVED ✅
## Gemini Review — Round 2 → VERDICT: APPROVED ✅
## Mercury Review — Round 2 → VERDICT: APPROVED ✅
VERDICT: APPROVED — unanimous after 2 rounds
Required:
npm install -g acpx@latestbrew install jq / apt install jqOptional:
These are the reviewer backends you can use. Mix and match — pick 2-4 for a useful review panel.
These have native Agent Client Protocol support. Install the CLI, and acpx handles the rest.
| Agent name | Model | Install |
|---|---|---|
codex | OpenAI Codex | npm install -g @openai/codex + OPENAI_API_KEY |
gemini | Google Gemini 2.x/3.x | npm install -g @google/gemini-cli + gemini auth + GEMINI_API_KEY ¹ |
claude | Claude (Opus/Sonnet) | Already installed — you're running it now |
kimi | Kimi (Moonshot AI) | See Kimi CLI docs |
kiro | Kiro (AWS) | See Kiro docs |
qwen | Qwen Code | See Qwen Code docs |
cursor | Cursor | Install Cursor IDE |
copilot | GitHub Copilot | gh extension install github/gh-copilot |
opencode | OpenCode (default model) | npm install -g opencode-ai |
kilocode | Kilocode | npx @kilocode/cli |
droid | Factory Droid | See acpx docs |
iflow | iFlow | See acpx docs |
pi | Pi Coding Agent | See acpx docs |
openclaw | OpenClaw | See acpx docs |
¹ Gemini note: The Gemini CLI's stored OAuth works for direct CLI use but not for acpx's non-interactive subprocess mode. You need a separate API key. Get one free at aistudio.google.com/apikey, then add it to
~/.claude/settings.json:"env": { "GEMINI_API_KEY": "AIza..." }Antigravity CLI transition (June 18, 2026): Google is migrating Google One and free-tier Gemini CLI users to the new Antigravity CLI. The
GEMINI_API_KEYpath used by this plugin is unaffected — AI Studio API keys bill against AI Studio quota, not Google One OAuth tiers, so thegeminireviewer continues to work. Paid Gemini Code Assist Standard/Enterprise users also keep thegeminiCLI as-is. A dedicatedantigravityreviewer will be added once acpx ships native ACP support for it.
Claude note: Using
claudeas a reviewer means Claude reviewing its own plan — useful for a fresh-context skeptical read, but not truly independent. For independent perspectives, use non-Claude agents.invoke-acpx.shautomatically handles the nested-session guard (CLAUDECODE) required to run Claude as a subprocess.
npx claudepluginhub strml/cc-debateDetects plugin marketplace updates at session start and flags security-relevant changes for review
Language-agnostic codemap generator for Claude Code. Scans your codebase and generates token-lean architecture maps that load into context at session start.
iTerm2 tab highlighting for Claude Code permission requests — orange tab + dock bounce on permission, auto-reset on focus
Automatically record browser demo videos of pull requests using agent-browser
Persistent cross-session memory for Claude Code. Hooks extract learnings automatically; Sonnet + Opus maintain MEMORY.md files per project and globally.
Give your AI a memory — mine projects and conversations into a searchable palace. 33 MCP tools, auto-save hooks, and guided setup.
Comprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Intelligent draw.io diagramming plugin with AI-powered diagram generation, multi-platform embedding (GitHub, Confluence, Azure DevOps, Notion, Teams, Harness), conditional formatting, live data binding, and MCP server integration for programmatic diagram creation and management.
Persistent file-based planning for AI coding agents. Crash-proof markdown plans (task_plan.md, findings.md, progress.md) that survive context loss and /clear, with an opt-in completion gate and multi-agent shared state. Manus-style. Works with Claude Code, Codex CLI, Cursor, Kiro, OpenCode and 60+ agents via the SKILL.md standard. Includes Arabic, German, Spanish, and Chinese (Simplified and Traditional).
Complete creative writing suite with 10 specialized agents covering the full writing process: research gathering, character development, story architecture, world-building, dialogue coaching, editing/review, outlining, content strategy, believability auditing, and prose style/voice analysis. Includes genre-specific guides, templates, and quality checklists.
TypeScript/JavaScript full-stack development with NestJS, React, and React Native