browser-automation

Unified productivity plugin for Claude Code providing comprehensive development tools and automated workflows.

Installation

Add this plugin to Claude Code:

/plugin install skill-set

That's it! All features are now available in a single unified plugin.

Features

Git Workflow Automation

Commands:

• •/skill-set:git:commit - Create context-aware git commits
• •/skill-set:git:push - Push changes to remote (auto-commits if needed)
• •/skill-set:git:pr - Create pull requests (auto-push and commit if needed)

Capabilities:

• •Language-aware commit messages (adapts to your project's language)
• •Automatic ticket number extraction from branch names
• •Context-aware PR descriptions with full change summaries
• •Git history analysis for consistent messaging

Code Context Understanding

Find and read official documentation for external libraries and frameworks using Context7.

Use when: Understanding external libraries, frameworks, or dependencies.

Features:

• •Official documentation lookup via Context7
• •Multiple search term strategies for finding library docs
• •Version-specific authoritative documentation
• •Best practices and patterns from official sources

Peer LLM Consulting

Execute peer reviews from other LLM tools (Gemini, Codex, Claude) in parallel and synthesize actionable insights.

Command:

/skill-set:consulting:review

Features:

• •Dynamic CLI selection with auto-detection
• •Parallel LLM execution for faster feedback
• •Synthesized reports with actionable insights

Ralph Loop

Plan and execute implementation work with fresh context per iteration via Task subagents.

Commands:

/skill-set:ralph:plan
/skill-set:ralph:plan docs/plans/feature-design.md
/skill-set:ralph:execute
/skill-set:ralph:execute tmp/ralph/2026-03-06-1430/plan.md

Features:

• •Two modes: PLANNING (generate plans) and BUILDING (implement from plans)
• •Fresh subagent per iteration (no context rot)
• •Plan file on disk as single source of truth
• •Git commit-based progress tracking
• •Auto-planning fallback (execute without a plan → auto-creates one)
• •DONE condition negotiation before loop starts

PR Review Feedback Processing

Interactive PR review processing from any source (human, CodeRabbit, Codex, Claude, other bots) with severity classification and verified completion workflow.

Command:

/skill-set:pr-review:fix

Features:

• •Processes review comments from any source (human reviewers, AI tools, bots)
• •Automatic severity-based classification (CRITICAL, MAJOR, MINOR)
• •Interactive issue discussion before applying changes
• •Verified completion workflow with mandatory steps

Skill Creation Guide

Comprehensive guide for creating effective Claude skills with structured workflow and testing methodology.

Writing Clear Prose

Guides writing and revision of explanatory text, persuasive proposals, and technical documents with 4 core principles.

Guarding Agent Directives

Guards agent directive files (CLAUDE.md, AGENTS.md) against bloat by verifying additions through strict criteria.

Session Initialization

Automatically establishes workflows at session start to ensure proper skill usage.

Usage

Skills are automatically available after installing the plugin. Claude will use them when relevant to your task.

Slash Commands

/skill-set:git:commit
/skill-set:git:push
/skill-set:git:pr
/skill-set:ralph:plan
/skill-set:ralph:execute
/skill-set:pr-review:fix
/skill-set:consulting:review
/skill-set:plan:grill
/skill-set:code:zoom-out

Project Structure

plugins/
└── skill-set/
    ├── .claude-plugin/
    │   └── plugin.json
    ├── .mcp.json
    ├── commands/
    │   ├── git/                    # commit, push, pr
    │   ├── ralph/                  # plan, execute
    │   ├── pr-review/              # fix
    │   └── consulting/             # review
    ├── skills/
    │   ├── managing-git-workflow/
    │   ├── understanding-code-context/
    │   ├── consulting-peer-llms/
    │   ├── ralph/
    │   ├── creating-skills/
    │   ├── writing-clear-prose/
    │   ├── guarding-agent-directives/
    │   └── using-skill-set/
    ├── agents/
    │   └── pr-review-feedback.md
    └── hooks/
        └── hooks.json

Design Philosophy

• •Progressive disclosure: Core functionality in SKILL.md, details in reference files
• •Context-aware: Adapts to your project's language and conventions
• •Token efficient: Reads only necessary code using symbolic tools
• •Namespaced: Commands organized to prevent collisions
• •Unified: Single installation, consistent experience

Requirements

• •Claude Code (latest version recommended)
• •Git (for git workflow features)
• •Peer LLM CLIs (for consulting features, optional):

- gemini-cli for Gemini - codex for Codex - Additional CLIs detected automatically

Contributing

Contributions welcome! Please see AGENTS.md for development guidelines and best practices.

License

MIT

Acknowledgements

The following skills were inspired by mattpocock/skills (MIT license):

• •grilling-plans ← grill-me / grill-with-docs
• •building-shared-vocabulary ← grill-with-docs (CONTEXT.md / ADR pattern)
• •zooming-out-on-code ← zoom-out
• •improving-architecture ← improve-codebase-architecture
• •developing-test-first (Horizontal Slicing section) ← tdd

Changelog

See CHANGELOG.md for version history and migration guides.