claude-code-plugin

Review codebase architecture for deepening opportunities — modules where the interface is nearly as complex as the implementation. Uses depth-as-leverage model and the deletion test. Use when the codebase feels tangled, after a burst of fast development, or when asked to review architecture.

Ultra-compressed communication. Drops filler, hedging, and pleasantries while keeping technical terms exact. ~75% token reduction. Use when user says "caveman", "caveman mode", "be terse", or "stop being chatty".

Structured debugging as a state machine: reproduce, minimize, hypothesize, instrument, fix, regression-test. Use when something is broken, a bug needs fixing, tests are failing unexpectedly, or behavior doesn't match expectations.

Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me".

Test-driven development with vertical slices. Enforces behavior-focused tests through public interfaces, one red-green-refactor cycle at a time. Use when writing tests, adding test coverage, or building features test-first.

Convert markdown to portable, shareable HTML documents with syntax highlighting and optional tabbed navigation. Use when users say "publish", "make HTML", "make this shareable", "create a document for the team", or want to share output with non-technical audiences.

Generates publication-quality illustrations using a 5-agent pipeline inspired by PaperBanana (academic illustration framework). Orchestrates specialized agents: Retriever, Planner, Stylist, Visualizer, and Critic. Use when users need high-quality, iterative image generation with style consistency and quality review.

Pedagogical code walkthrough that teaches developers about their own code. Explains choices, tradeoffs, alternatives, and refactoring opportunities. Use when users say "teach me", "explain my changes", "walk me through", "what did I build", or want to understand their code at a deeper level. Works on uncommitted changes, last commit, unpushed commits, or a PR.

Detects common failure patterns in proposals, architectures, plans, and decisions. Use when reviewing any plan to identify red flags before they become problems. Covers architecture, timeline, team, process, and technology anti-patterns. Works on any domain — not just code.

Identifies and stress-tests implicit assumptions in plans, proposals, and decisions. Use when validating roadmaps, reviewing architecture proposals, assessing timelines, or challenging any plan where hidden assumptions could cause failure. Works on any domain — code, business, strategy, operations.

Reviews code for logic flaws, edge cases, performance bottlenecks, security vulnerabilities, maintainability concerns, architecture best practices, and adherence to local patterns and CLAUDE.md guidelines. Use when users ask for code review or analysis, including reviewing uncommitted changes, the last local commit, unpushed code, or a specific GitHub pull request.

Guides users through defining features via adaptive discussion and generates structured feature 1-pagers. Use when users want to create a well-documented feature request or need help articulating a feature idea. The discussion is conversational, not a form - follow threads, ask follow-ups, and refine until the user is satisfied.

Gets Gemini's independent opinion on any file, topic, or discussion. Use when users say "ask gemini", "gemini opinion", or "what does gemini think". Provides a second perspective from a different AI model with Claude's commentary. Requires GEMINI_API_KEY environment variable to be set.

Dual code review: gets independent reviews from both Claude and Gemini API, then synthesizes agreements, unique findings, and a verdict. Use when users say "gemini review", "dual review", or want a second-perspective code review. Requires GEMINI_API_KEY environment variable to be set.

Generates images using Google Gemini's image generation models (Nano Banana). Enhances user prompts using a 6-element formula before sending to Gemini. Supports text-to-image and image-to-image generation. Use when users want to create, generate, or modify images with AI.

Creates GitHub issues and adds them to the ExamJam V.Next 25 project (project 19) with custom field values. Handles the complete workflow of gathering issue details, fetching project field options dynamically, creating the issue, adding it to the project, and setting Priority, Type, Initiative, and Status fields.

This skill should be used when the user asks to "create a PR", "open a pull request", "write a PR description", "push and create PR", or needs to generate a best-in-class pull request description based on commits. Provides industry-standard PR description formatting following conventional commits, Microsoft Engineering Playbook, and HackerOne best practices.

Synthesizes validation analysis into a structured 8-section report with clear verdict (GOOD / NEEDS MAJOR WORK / BAD), strengths, critical flaws, blindspots, and actionable path forward. Use after assumption-challenger and antipattern-detector have produced their analyses, or standalone for quick validations. Produces the final deliverable for /validate.