AI Coding Configuration

Curated commands, agents, and rules for Claude Code, Cursor, Windsurf, and Cline.

What This Is

A shared configuration that works across AI coding tools. Commands automate workflows (PR handling, debugging, session management). Agents specialize in specific review types (security, performance, UX). Rules encode your coding standards so AI follows your patterns.

Quick Start

One-line install

Run this in your terminal — it auto-detects your tools (Claude Code, Cursor, etc.) and sets everything up:

curl -fsSL https://raw.githubusercontent.com/TechNickAI/ai-coding-config/main/scripts/bootstrap.sh | bash

Then configure for your project:

/ai-coding-config

Manual install (Claude Code)

If you prefer to install manually, run these in Claude Code:

/plugin marketplace add https://github.com/TechNickAI/ai-coding-config

/plugin install ai-coding-config

Todo Persistence Across Compaction

The problem: Claude Code's context compaction summarizes conversation history to stay within token limits. When this happens, your todo list vanishes - you lose track of what you were working on.

The solution: This plugin automatically saves todos to disk via hooks. After compaction, restore them:

cat ~/.claude/projects/$(echo $PWD | sed 's|/|-|g')/todos.md

Then use TodoWrite to restore your progress. Your todos survive compaction, session restarts, and even --resume across days.

Example Usage

# Triage PR bot comments (fixes real issues, declines nitpicks)
/address-pr-comments

# Load coding standards for current task
/load-rules

# Autonomous task execution: describe what you want, get a PR
/autotask "add user settings page with dark mode toggle"

What's Included

Type	Count	Purpose
Commands	18	Automate workflows
Agents	24	Specialized assistants
Skills	6	Autonomous capabilities
Rules	33	Coding standards
Personalities	7	Communication styles

---

Commands

Slash commands that automate real workflows. Type /command-name to invoke.

Autonomous Development

/autotask "description" - Describe what you want, get a PR. Creates branch, implements, writes tests, handles bot feedback.

/autotask "add user settings page with dark mode toggle"

/troubleshoot - Connects to Sentry/HoneyBadger, analyzes errors, fixes bugs in parallel worktrees, submits PRs with root cause analysis.

/verify-fix - Confirms fixes actually work before claiming success. Runs tests, checks live behavior.

PR & Code Review

/address-pr-comments - Triages bot comments: fixes real issues, declines nitpicks, iterates until merge-ready.

/multi-review - Runs multiple specialized reviewers in parallel. Security, logic, performance, style - all at once.

Context & Session Management

/session save|resume|list - Save your context, decisions, and progress. Resume exactly where you left off - even in a new conversation.

/session save "auth-refactor"    # Save current session
/session resume                   # Resume where you left off

/load-rules - Loads relevant coding standards for your current task. Working on React? Loads React patterns. Writing tests? Testing standards.

/handoff-context - Generate context handoff for new sessions.

Project Setup

/ai-coding-config - Interactive setup for new projects. Detects your stack, installs relevant rules.

/repo-tooling - Set up linting, formatting, CI/CD based on detected language.

/setup-environment - Initialize dev environment for git worktrees.

Other Commands

• /cleanup-worktree - Clean up worktrees after PR merge
• /generate-AGENTS-file - Generate AGENTS.md for AI context
• /generate-llms-txt - Generate llms.txt for LLM navigation
• /knowledge - AI Product Manager - maintain living product understanding
• /personality-change - Switch AI communication style
• /product-intel - Competitive intelligence research

---

Agents

24 specialized assistants that Claude Code invokes automatically based on context.

Building Features

Agent	Purpose
autonomous-developer	Implements features end-to-end following your patterns
test-engineer	Writes comprehensive test coverage
test-runner	Runs tests with terse, context-efficient output

Debugging

Agent	Purpose
debugger	Root cause analysis through systematic investigation

Code Review - Correctness

Agent	Purpose
logic-reviewer	Bug and logic error detection
error-handling-reviewer	Silent failures and try-catch patterns
security-reviewer	Injection, auth, OWASP top 10
robustness-reviewer	Production readiness and resilience

Code Review - Performance

Agent	Purpose
performance-reviewer	N+1 queries, algorithmic complexity
simplifier	Reduce complexity, preserve functionality

Code Review - UX

| Agent