Use when asked for an architecture blueprint or explicitly asked to run the code-architect subagent.

code-audit-vigil

Analyze and map code structure, data flow, and dependencies in legacy codebases. Use for understanding call stacks, dependency trees, side effects, risk profiles, and generating Mermaid diagrams. Ideal for legacy codebases with zero tests.

Detect and refactor code duplication with PMD CPD. TRIGGERS - code clones, DRY violations, duplicate code.

Multi-language code coverage analysis, reporting, and quality gate enforcement

Standardized, hypothesis-driven debug investigation workflow for unexpected behavior, regressions, incidents, flaky tests, or production issues. Use to plan instrumentation, reproduce with decisive signals, compare runs, and narrow root cause without assuming any vendor or infrastructure.

Use when explaining code, concepts, or system behavior to a specific audience level - provides a structured explanation workflow with depth control and validation steps.

Use when asked to trace existing codepaths or explicitly asked to run the code-explorer subagent.

Use when generating boilerplate code, __init__.py files, or test scaffolds. Provides scripts that generate consistent, convention-following code.

code-guru

Language-specific coding guidelines. Use when implementing code in Python or other supported languages.

Migrate legacy code to modern frameworks, languages, and patterns with automated refactoring and testing

Use the VSCode CLI to open files at specific line and column positions when explaining code, debugging, pointing to TODOs, or guiding users to specific locations in files. Automatically invoked when Claude needs to show users exactly where in a file to look.

Use when asked to simplify recently changed code without changing behavior or explicitly asked to run the code-simplifier subagent.

Enforce test-driven development (RED-GREEN-REFACTOR). Use when implementing features, fixing bugs, or changing behavior - write failing test first, then minimal code to pass.

QA engineer and test automation specialist with deep expertise in Flutter testing. Use for designing test strategies, writing unit/widget/integration tests, improving test coverage, and ensuring code reliability.

Transfer code between files with line-based precision. Use when users request copying code from one location to another, moving functions or classes between files, extracting code blocks, or inserting code at specific line numbers.

Validate code examples across the 4-Layer Teaching Method with intelligent strategy selection. Use when validating Python/Node/Rust code in book chapters. NOT for production deployment testing.

Analyze a GitHub codebase to create comprehensive architecture documentation including ASCII diagrams, component relationships, data flow, hosting infrastructure, and file structure assessment.

Expert in codebase archaeology and documentation retrieval. Use this when starting new project phases, refactoring large modules, or integrating new frameworks.

Deep codebase exploration using semantic search and relationship mapping. Use when you need to understand the current codebase.

Unified dead code detection, AST-based duplicate detection, and cleanup tool combining Legacy Tech Remover + Comprehensive Auditor + Architectural Cleanup with Knip, depcheck, TypeScript, Vue-specific analysis, file size/token limit detection, and AST-powered refactoring suggestions. Use for dead code, duplicates, oversized files, and architectural issues.

>

Explores codebase and writes structured analysis documents. Spawned by map-codebase with a focus area.

Monitor and refactor large files into smaller, AI-friendly modules. Use when user asks to check file sizes, split large files, or organize the codebase. Ensures tests pass before and after refactoring.

Comprehensive pre-merge testing for CodeGeass. Tests PR changes FIRST, then runs regression tests on all CLI commands and Dashboard API endpoints.

>

>

Analyze code for style violations, pattern compliance, and quality issues

Execute development tasks using OpenAI Codex CLI for code generation, refactoring, feature implementation, and bug fixes. Use when the user asks to create code, add features, refactor, fix bugs, or generate tests. Requires Codex CLI installed.

Create or edit Codex execpolicy .rules files (allow/prompt/forbid commands, define prefix_rule patterns, add match/not_match tests) and validate them with codex execpolicy check. Use when a user mentions Codex rules, execpolicy, command policies, allowlists/denylists, or controlling which commands Codex can run, and when scope (global vs project) must be clarified.

Invoke OpenAI Codex CLI with the correct configuration for autonomous code generation. Use when you need Codex to generate or regenerate code files, implement features, or execute multi-step coding tasks.

Use Codex CLI in full-auto mode to fix issues iteratively until tests pass. Autonomous debugging and test-fixing loop with sandbox safety.

Run the Codex Readiness integration test. Use when you need an end-to-end agentic loop with build/test scoring.

Resume a previous Codex conversation session.

Use at the start of a Codex session (especially sandboxed) to run `scripts/codex-sandbox-preflight.sh` and interpret network + writable_roots constraints.

Investigate the @openai/codex-sdk API by first inspecting installed .d.ts typings in node_modules, then cloning and searching the github.com/openai/codex repo for source/docs. Use for questions about Codex SDK API surface, types, response formats, threads/runs, or debugging codex-sdk behavior.

Generate nested AGENTS.md coding guidelines per module (monorepo-aware), detect languages/tooling, ask architecture preferences, and set up missing formatters/linters (Spotless for JVM). Use when the user wants module-scoped AGENTS.md coding guidelines or to set up missing formatters/linters.

Verify changes follow nearest-scoped AGENTS.md rules: group changed files by nested scope, auto-fix formatting, run lint/tests, and report violations. Use when the user wants scoped compliance checks for changed files.

Comprehensive cognitive mode management skill for the VERILINGUA x VERIX x DSPy x GlobalMOO integration. Enables automatic mode selection, frame configuration, VERIX epistemic notation, and GlobalMOO optimization. Use this skill when configuring AI behavior for specific task types, optimizing prompt engineering, or ensuring epistemic consistency in responses.

Continuous business health monitoring across 13 business domains. Scores domain health, generates periodic digests, detects anomalies, simulates decisions, and provides early warning signals for problems.

Master Cohere v2 Chat API with Python, specializing in entity extraction using JSON Schema mode for structured outputs. Use when extracting entities from text, building data extraction pipelines, implementing NER systems, or requiring validated JSON responses from LLMs.

Generate personalized cold email sequences (7-14 emails) with A/B test subject lines, follow-up timing recommendations, and integrated social proof. Creates multi-touch campaigns optimized for response rates. Use when users need outbound email campaigns, sales sequences, or lead generation emails.

Create 25-35 second cold opens that hook listeners by dropping them into a specific moment that encapsulates the episode's themes. Uses narrative snippets method (story arcs with beats) combined with Colin & Samir's rearrangement technique.

Generate cold email templates, LinkedIn messages, follow-up sequences, and value propositions tailored to specific industries that actually get responses and book discovery calls.

Analyze staged files, group by category, and generate conventional commit specs as JSON.

Detects color contrast violations that make text unreadable and provides WCAG-compliant fixes. Uses both mathematical contrast ratio analysis and perceptual evaluation via vision capabilities.

>

Executes safe commands and captures real output to extend command-verify functionality. Only runs when explicitly requested. Use when asked to actually run commands, execute commands, test if commands work, or verify commands actually execute.