Skip to content

QUICKSTART.md

Latest commit

 

History

History
138 lines (103 loc) · 3.88 KB

QUICKSTART.md

File metadata and controls

138 lines (103 loc) · 3.88 KB

Warden - Quick Start (5 Minutes)

Get started with Warden in 5 minutes or less.

Prerequisites

  • GitHub CLI (gh) installed and authenticated
  • AI coding assistant (Claude Code, GitHub Copilot, or Cursor)
  • Warden repository in your workspace/directory

How It Works

  1. Open the Warden repository in your editor/workspace
  2. AI reads the instructions (CLAUDE.md, .cursorrules, or copilot-instructions.md)
  3. Say "Run Warden" explicitly
  4. AI executes the workflow automatically

Basic Usage

1. Review Your PRs (Default Settings)

"Run the Warden skill"
"Execute Warden on my open PRs"

This runs with safe defaults:

  • Standard review depth (1 generalist reviewer)
  • Tests affected packages only
  • Balanced fix strategy
  • Auto-commit on success

2. Preview Without Fixes (Dry Run)

"Run Warden in dry-run mode"
"Execute Warden without making changes to preview issues"

3. Security-Focused Review

"Run Warden with security and testing reviewers, use full test suite"

→ Uses: --reviewers security,testing --test-strategy full

4. Fast Review (Documentation PRs)

"Run Warden and skip tests for doc-only changes"

→ Uses: --test-strategy none

Common Scenarios

Scenario 1: Quick Daily Review

"Run Warden quietly with affected tests only"

→ Uses: --quiet --test-strategy affected

Scenario 2: Production Code Review

"Run Warden with conservative fixes, full tests, and review before pushing"

→ Uses: --fix-strategy conservative --test-strategy full --require-review-before-push

Scenario 3: Fix Specific PR

"Run Warden on PR #123"
"Execute Warden to fix PR #456"

Scenario 4: Comprehensive Audit

"Run Warden with security, performance, and architecture reviewers, full test suite, conservative fixes"

→ Uses: --reviewers security,performance,architecture --test-strategy full --fix-strategy conservative

What Warden Does

  1. Discovers PRs - Finds your open pull requests
  2. Analyzes Three Sources:
    • CI failures (test failures, build errors)
    • Review comments (requested changes)
    • Code quality (security, performance issues)
  3. Prioritizes - Sorts by severity (Critical → High → Medium → Low)
  4. Asks Permission - Shows you what it found, asks what to fix
  5. Fixes & Validates:
    • Applies fixes
    • Runs: Build → Lint → Format → Test
    • Only commits if ALL validations pass
  6. Pushes Changes - Updates the PR with fixes
  7. Reports - Shows summary of what was fixed

Key Features

  • ✅ Works on existing PRs - Not creating new ones
  • ✅ Discovers commands from your repo - Reads CLAUDE.md, CI configs for build/test commands
  • ✅ Validates before pushing - Build + Lint + Test must pass
  • ✅ Isolated workspaces - Never touches your working directory
  • ✅ Incremental fixes - Fix by severity tier, rollback if tests fail

Parameters

You can specify parameters in natural language. The AI understands:

Review depth:

  • "Use security and performance reviewers"
  • "Run comprehensive review with all specialists"

Test strategy:

  • "Use full test suite"
  • "Test affected packages only"
  • "Skip tests"

Fix strategy:

  • "Use conservative fixes"
  • "Be aggressive with fixes"
  • "Balanced approach"

Safety:

  • "Require review before pushing"
  • "Create rollback branch"
  • "Protect main and master branches"

See CONFIGURATION.md for workspace setup and PARAMETERS.md for all options (25 core + 19 advanced).

Next Steps

Remember

Always say "Run Warden" or "Execute the Warden skill" so the AI knows to use this specific workflow!