BeClaude

workflow-tools

New
1Community RegistryGeneralby AncpLua

Multi-agent workflow orchestration: /fix (unified fix pipeline), /red-blue-review (adversarial security), /tournament (competitive coding), /mega-swarm (parallel audit), /deep-think (extended reasoning), /batch-implement (parallel implementation).

Community PluginView Source

Overview

Claude Code plugin marketplace — an agent-teams-aware advanced software-engineering base set. 9 plugins for parallel agent orchestration, parallel code review, cognitive guardrails, behavior-first test quality, plugin/skill evaluation, and code-elegance work.

What this does

Built around Claude Code's multi-agent capabilities — from one-shot subagent fan-out (exodia spawns up to 12 workers) to the experimental agent-teams feature (council reconciles an Opus team against a /deep-research report via the Teams API). Each phase is gated — work only advances when the gate passes. No manual babysitting.

Plugins

9 plugins, 22 commands, 14 skills, 21 agents:

PluginWhat it does in plain language
exodiaSummons up to 12 AI workers that tackle different parts of your code simultaneously. One finds problems, another fixes them, another reviews the fix — all at the same time
councilCouncil for complex tasks. Opus captain decomposes and runs the /deep-research dynamic workflow, an Opus synthesizer reasons over it, an Opus clarity agent gap-checks it live, and an Opus janitor flags bloat (Teams API)
elegance-pipelineMulti-agent pipeline that scores code elegance and refactors the weakest files through gated stages — plus a standalone simplifier for quick diff cleanups
metacognitive-guardWatches the AI while it works. If it's about to cut corners, guess instead of verify, or say "done" when it isn't — this blocks it before the mistake happens
feature-devA guided process for building new things: understand what exists, design the plan, build it, review it. No skipping steps
mutation-minded-testingJudges whether your tests actually catch bugs, not just cover lines. Four agents review architecture, test quality, and expressiveness
cc-plugin-evalEvaluates and scores Claude Code plugins and skills: static analysis, token budgeting, and benchmarking across every component type
html-effectivenessRoutes output into a single self-contained HTML file — dashboards, reports, code reviews, slide decks — instead of a wall of markdown
nuget-opensrcFetches the exact source a NuGet package was built from, so the agent greps real code instead of guessing API shapes

How does this work without failing?

Every step is a gate. Work only moves forward if the gate says PROCEED. If it says HALT — the work stops, gets diagnosed, and gets fixed before anything else continues. No hoping. No skipping. No "it probably works."

Install

Add the marketplace, then install plugins individually:

bash
# Add the marketplace
/plugin marketplace add ANcpLua/ancplua-claude-plugins

# Install plugins you want
/plugin install exodia@ancplua-claude-plugins
/plugin install council@ancplua-claude-plugins
/plugin install metacognitive-guard@ancplua-claude-plugins
/plugin install elegance-pipeline@ancplua-claude-plugins
/plugin install cc-plugin-eval@ancplua-claude-plugins

Technical details

9 plugins, 22 commands, 14 skills, 21 agents.

Tri-AI review system: Claude, Copilot, and CodeRabbit review PRs independently.

GitHub review automation

Codex review lives in .github/workflows/codex-code-review.yml.

  • Set the OPENAI_API_KEY repository secret to enable the workflow
  • Codex runs in a read-only sandbox with drop-sudo
  • The workflow posts a formal PR review from structured Codex output
  • Self-review is blocked when a PR only changes Codex review automation files
text
plugins/
├── exodia/                  # parallel agent orchestration (9 commands + 2 skills)
├── council/                 # Opus council + /deep-research workflow (Teams API)
├── elegance-pipeline/       # code elegance scoring + gated refactoring + standalone simplifier
├── metacognitive-guard/     # quality gates + commit integrity + CI verification
├── feature-dev/             # guided feature development + code review
├── mutation-minded-testing/ # behavior-first test quality (4 agents)
├── cc-plugin-eval/          # plugin & skill evaluator (static analysis + benchmarking)
├── html-effectiveness/      # route output to a self-contained HTML file
└── nuget-opensrc/           # commit-pinned NuGet → source fetcher

Links

License

MIT

Install & Usage

1
Create the skills directory
mkdir -p .claude/skills
2
Download the skill file
mkdir -p .claude/skills && curl -o .claude/skills/workflow-tools.md https://raw.githubusercontent.com/ANcpLua/ancplua-claude-plugins/main/SKILL.md
3
Invoke in Claude Code
/workflow-tools
View source on GitHub
securitycode-reviewagent

Frequently Asked Questions

What is workflow-tools?

Multi-agent workflow orchestration: /fix (unified fix pipeline), /red-blue-review (adversarial security), /tournament (competitive coding), /mega-swarm (parallel audit), /deep-think (extended reasoning), /batch-implement (parallel implementation).

How to install workflow-tools?

To install workflow-tools, create the .claude/skills directory in your project, then run the curl command to download the skill file. Once installed, invoke it in Claude Code with /workflow-tools.

What is workflow-tools best for?

workflow-tools is a community categorized under General. It is designed for: security, code-review, agent. Created by AncpLua.