Actual STATE - work in progress; Please add your changes as PR.

Install the tooling into any existing project with a single command:

Option A — curl (no clone needed):

curl -fsSL https://raw.githubusercontent.com/JoeRu/vibe-coding-claude-template/main/setup.sh | bash -s -- /path/to/your/project

Option B — from a local clone:

git clone https://github.com/JoeRu/vibe-coding-claude-template.git
cd vibe-coding-claude-template
./setup.sh /path/to/your/project

Install into the current directory:

./setup.sh        # or: curl ... | bash

Preview without making changes:

./setup.sh /path/to/your/project --dry-run

The script is idempotent — re-running is safe. Existing files are backed up to <file>.bak before overwriting.

After installation, open your project in Claude Code and run /init_overview.

See the Usage Guide for step-by-step walkthroughs, examples for new and existing projects, and a full modifier reference.

This Project has the intend to add better results when using vibe-coding agents. The tendency is to keep lost of features, bugs, debts or security features. The XML and the Chapter for your CLAUDE.md or AGENT.md has the intend to make the use more structured.

All future .md files generated by the AI are now stored at ai-docs folder

All installable files live under template/. The installer (setup.sh) copies them into your target project:

template/
  .claude/commands/          ← slash commands for Claude Code
  .github/
    copilot-instructions.md  ← baseline Copilot instructions
    prompts/                 ← equivalent prompts for GitHub Copilot
  ai-docs/
    implementation-plan-template.xml
  CLAUDE-implementation-plan-chapter.md

Add CLAUDE-implementation-plan-chapter.md to your CLAUDE.md, AGENT.md or copilot-instructions.md. The easiest way is to use setup.sh (see Quick Start above). To do it manually:

1. Copy the contents of template/ into your project root.
2. Add the following lines to your agent file (CLAUDE.md / AGENT.md):

## Implementation Plan Workflow

**IMPORTANT:** Read `CLAUDE-implementation-plan-chapter.md` for the XML-based implementation plan workflow. All feature requests, bugs, and changes must be tracked in `ai-docs/overview-features-bugs.xml`.

On first encounter with a codebase, perform the initial analysis with /init_overview and generate ai-docs/overview.xml and ai-docs/overview-features-bugs.xml.

After updating CLAUDE.md and commands you need to restart Claude Code for changes to take effect.

There are several new slash commands enabled when adding this chapter to your AI behaviour. If you are not using Claude Code — e.g. GitHub Copilot — copy the prompts from template/.github/prompts/ instead of template/.claude/commands/.

• /bug <description>: Create a bug item with default priority HIGH.
• /feature <description>: Create a feature item with default priority MEDIUM.
• /refactor <description>: Create a refactoring item with default priority MEDIUM.
• /debt <description>: Create a tech-debt item with default priority LOW.

• /approve <ID> [ID...]: Approve items and generate branch names.
• /implement [ID...]: If a special ID is commited it's autoapproved, if none is commited - all Feature with APPROVED status will be implemented (max 5 in a row)
• /deny <ID> [reason]: Deny an item and archive it immediately.
• /status <ID>: Show current status, tasks, and dependencies.
• /list [filter]: List items by status, type, or priority.
• /archive: Archive eligible DONE/DENIED items and sync overview data.

• /security: Run a full security audit of the codebase.
• /security <area>: Run a focused security audit (e.g. auth, api).
• /security status: Show open concerns and security coverage.

• /plan: Show plan summary with next actionable items.
• /overview: Show architecture summary from overview.xml.
• /init: Run initial analysis and create both XML files.

• Single source of truth: Maintains active and archived work items in structured XML.
• Clear lifecycle states: Enforces PENDING → APPROVED → IN_PROGRESS → DONE or DENIED.
• Consistent ID and branch rules: Sequential IDs and standardized branch naming keep work traceable.
• Built-in security evaluation: Requires security impact assessment for relevant changes.
• Complexity-aware planning: Forces decomposition of XL items into manageable sub-items.
• Audit-ready changelog: Records each interaction's changes in a single entry.
• Archival with project baseline sync: Ensures completed work is reflected in overview.xml.