wizard

Guides phased software development with TDD, adversarial review, and GitHub issue tracking.

1|Updated May 16, 2026
One-click install
npx skills add https://github.com/ludotype/game-adventurertown --skill wizard-ludotype
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: wizard
Source: https://github.com/ludotype/game-adventurertown/tree/main/.claude/skills/wizard
Command: npx skills add https://github.com/ludotype/game-adventurertown --skill wizard-ludotype

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Complex features, bug fixes, and multi-file refactors often fail due to skipped planning, unverified assumptions, race conditions, and missing tests. This Skill enforces an architect-level development methodology that prevents these failure modes before code is written. ## Core Features & Use Cases - Phased Development Workflow: Eight structured phases covering planning, codebase exploration, TDD (RED/GREEN), implementation, test verification, documentation, pre-commit review, and PR quality gates. - Adversarial Self-Review: Built-in checklists and questions that force verification of edge cases, concurrency risks (TOCTOU), transaction side effects, and mutation-resistant test assertions. - GitHub Issue Integration: Creates and tracks GitHub issues with acceptance criteria as the source of truth for medium and complex tasks. - Use Case: When asked to fix a bug in a shared-state module, the Skill first maps all concurrent access paths, writes failing boundary tests, implements an atomic state transition, runs the affected test suite, and resolves all automated PR review findings before merge. ## Quick Start Use the wizard skill to implement this feature following TDD and the full phased development workflow.

Frequently Asked Questions about wizard

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I apply test-driven development to a complex feature?▼

Follow the RED-GREEN cycle: write failing tests for behavior that does not exist yet, confirm they fail for the right reason, then implement the minimal code to pass. Add boundary tests (0, 1, -1, null, empty) and assert specific values so mutations like changing > to >= are caught.

How do I prevent race conditions in shared state code?▼

Use atomic check-and-act operations instead of separate read-then-update steps, which create TOCTOU vulnerabilities. Lock the record, read state, act, then unlock, or use a single conditional UPDATE statement that only succeeds when the expected state still holds.

When should I create a GitHub issue before coding?▼

Create a GitHub issue for medium tasks (2-3 files) and complex tasks (4+ files or architectural impact). The issue holds acceptance criteria and serves as the source of truth throughout development, with progress comments and label updates at milestones.

What test suite should I run after making changes?▼

Match the suite to change scope: related tests only for single-file fixes under 20 lines, feature test suites for multi-file changes, and all affected modules for cross-cutting, database, or auth changes. Never commit with failing tests.

Why do error-handling updates disappear when a transaction fails?▼

Any change made inside a transaction is rolled back when an exception triggers rollback, including audit records and failure status updates. Persist error state outside the transaction in a catch block so it survives the rollback.