ai-playground

Read the current project's current and prior Claude Code or Codex sessions, extract every prompt, score five effort metrics 0-100 with reasons, estimate average tokens and time per session, and render a self-contained light-theme HTML report. Use when the user runs /effort-map or asks how much effort, thought, bug-fixing, UX tuning, architecture care, or copy-paste slop went into a project.

Find bugs that were encountered and solved in prior Claude Code or Codex sessions for the current project, then generate a self-contained searchable light-theme HTML bug report. Use when the user asks for a bug catalog, solved-problem history, prior-session bug report, recurring failure inventory, or a BUG to Solution report for the current repository.

Analyze a prompt with Edward de Bono's Six Thinking Hats and produce a polished, self-contained HTML report. Use when the user invokes /hats or $hats, requests Six Thinking Hats analysis, wants a decision reviewed from factual, emotional, cautious, optimistic, creative, and process perspectives, or asks for an illustrated HTML decision report.

Reads every Claude Code session in the user's config directory and renders a self-contained light-theme website with a GitHub-style contribution grid, visual metrics, and 5-10 written insights about the user's agent coding habits and what they are building. Use when the user runs /habit-tracker or asks about their coding habits, activity streak, contribution graph, what they have been building, or a recap of their Claude Code usage.

Rewrite poor git commit messages into a clear conventional-commit history. Reads the last N commits with their diffs, proposes a clean conventional message for each by reading what the code actually changed, then opens a light-theme review UI where the user approves all or each message individually before the history is rewritten. Use when the user runs /fix-git-history or asks to clean up, rewrite, normalize, or fix commit messages or git history into a conventional-commit format.

Audits a running web page for WCAG accessibility issues, captures a screenshot and annotates it with a numbered box over every problem, then renders a self-contained light-theme website with the score, the annotated screenshot, and a fix for each finding. Use when the user runs /accessibility-auditor or asks to audit, check, or report a site's accessibility, WCAG compliance, alt text, color contrast, labels, or screen-reader support.

Audit a CLAUDE.md for cruft and render a light-theme report website. Reads the file line by line and flags vague rules with no checkable criterion, low-signal lines that only restate what the model already does, overlapping rules said more than once, and internal contradictions, then scores the file with a BS-o-meter and grade. Can also diff two CLAUDE.md files and flag rules that tell the model opposite things. Use when the user runs /bs-claudemd or asks to review, audit, grade, lint, or clean up a CLAUDE.md / AGENTS.md / agent instructions file, or to compare two of them.

Find dead Java code by static reachability from entry points. Parses the sources, builds the method call graph, walks it from real entry points (main, Spring web/lifecycle/scheduler handlers, JUnit tests, framework callbacks), and renders a light-theme website with summary cards, a reachable-vs-dead chart, and a grouped list of every unreachable method and dead class. Use when the user runs /dead-code or asks to find dead, unused, or unreachable code, or which methods and classes are safe to delete.

Runs a real re-render benchmark on every React component in a project and renders a Lighthouse-style light-theme performance report. For each component it mounts the real code in jsdom inside a React Profiler, forces a fixed number of parent re-renders with stable props, and counts the renders React could have skipped (wasted renders) plus render durations, then scores each component 0-100. Use when the user runs /rerender or asks to measure, benchmark, or find React re-render / wasted-render / memoization performance problems.

Read active Prometheus alerts and propose concrete runbook remediation steps for each firing alert. Use when the user runs /prometheus-runbooks or asks to triage Prometheus or Alertmanager alerts, investigate firing alerts, explain what an alert means, or generate a runbook for an alert.

Build a function call-graph flamegraph website for a Java codebase. Parses the sources, extracts every method and the calls between them, and renders a light-theme site where the user picks any class or method and sees its call tree as a hand-drawn SVG flamegraph. Use when the user runs /flamegraph or asks for a call graph, call-graph flamegraph, or a visualization of which methods call which.

Attributes a frontend project's JavaScript bundle size to the imports that caused it, then renders a light-theme analytics website. Bundles the app with esbuild, reads the metafile, and reports how many minified KB each npm package and source module added. Use when the user runs /bundle-size or asks which import is the heaviest, what is bloating the bundle, or for a bundle-size / import-cost report.

Builds a branch graveyard for a git repository. Reads every local and remote branch, records who last touched it and when, computes age, ahead/behind, and whether it is merged into the trunk, then renders a light-theme website that shows the stale branches as headstones and lists the ones that are safe to delete. Use when the user runs /branch-tombs or asks about stale, dead, abandoned, or old branches, branch cleanup, or which branches are safe to prune.

Generate ephemeral catching tests on the current diff, run them, and report behavior regressions. Tests are never committed. Works across Java, Kotlin, Scala, Python, and Node.js stacks.

Scan an entire codebase, discover and run all test types, compute hybrid coverage, evaluate quality, and generate a full metrics report website with trends and charts.

Produce a flashy light-theme website that recaps the past week of work by combining the current git repository's history with the user's Google Calendar events. Use when the user runs /weekly-review or asks for a weekly summary, recap, or report of their work.

Reads a codebase and discovers its database schema wherever it is defined — Liquibase changelogs, raw SQL run through JDBC, and Hibernate/JPA entities — merges them into one model, and renders a self-contained light-theme website with three tabs (a per-table data dictionary, a hand-drawn ER diagram, and an in-browser SQL query console). Use when the user runs /data-dict or asks to document, map, visualize, or generate a data dictionary for a project's database schema.

Detect whether a target web project is a React app on Node or Bun, run it, hunt for CSS / functional / render bugs with Playwright, and record an optimized video of each bug. Use when the user runs /bug-recording or asks to find and record UI bugs in an app.

Render a self-contained light-theme website with three tabs (bugs found, bug videos, reproduction steps) from a bug-recording output directory. Use when the user runs /bug-report or asks to turn recorded bug findings into a report site.

Builds a knowledge-risk heatmap of a git repository. Runs git blame across tracked files, finds the ones owned by a single person, computes a bus factor per file, directory, and author, and renders a light-theme website that shows where knowledge is concentrated. Use when the user runs /bus-factor or asks where the project's knowledge risk, single points of failure, or single-owner code lives.

Author and verify engineering rules as a single HTML contract. Use when the user runs /html-as-contract to review, accept, or suggest rules, or /html-as-contract check to scan code and regenerate the compliance report tab. Callable by a human or by another agent.

Lint a codebase on code quality, best practices, design principles, build and test status, per-test speed (slow at or above 5s), cyclomatic complexity, and expressive naming, then render the results as a modern web report. Use when the user runs /lint or /lint-site, or asks to lint, score, or audit the quality of a project.

Reads reverse-postmortem.md and renders a self-contained static site (reverse-postmortem-site/index.html) with a risk dashboard, per-incident cards, causal-chain visuals, and action-item checklists. No build step, no external dependencies.

Scans the codebase, predicts the most likely future incidents, and writes each as a full past-tense postmortem (timeline, root cause at file:line, detection gap, preventive action items) into reverse-postmortem.md.

Run themed RPG learning sessions from `/rpg-learn` requests. Use when the user wants a visual browser game or terminal fallback with quests, cards, XP, score, inventory, failures, boss fights, and safe shell command quests for Algorithms, Data Structures, Generative AI, Machine Learning, SRE/DevOps, or Management.

Use when a user wants a funny local pain map of a repository showing complex files, churn hotspots, stale code, ownership risk, weak tests, bug-heavy areas, and maintainability risk through /roast or /roast-map.

Use this skill to generate well-branded interfaces and assets for the Diego Pacheco engineering portfolio — `https://diegopacheco.github.io/`. Suitable for production code, throwaway prototypes, additional portfolio pages, book/talk microsites, and slide decks that should match the existing look. Contains essential design guidelines, colors, type, fonts, assets, and a UI kit of React components.

A minimal placeholder skill that greets the user from the claude-skill-marketplace POC

Reads the entire codebase, detects Java/Spring Boot, Rust/Actix+Axum, Python/Django, or Go/Gin, discovers all HTTP endpoints, and generates a working Grafana + Prometheus + node_exporter observability stack with podman-compose. Dashboard includes machine metrics (CPU, memory, disk, I/O, network) and application metrics (request counters, error counters, latency p50/p70/p90/p95/p99/p99.9 per endpoint).

Scans a project, auto-detects language/framework, and generates infrastructure files (Containerfile, K8s, Helm, OpenTofu, Ansible) based on user choices.

Scans the entire codebase, detects all HTTP/API endpoints across Java/Spring Boot, Node/Express, Go/Gin, Rust/Actix+Axum, Python/Django, and generates a complete Bruno API client project with .bru files, sample requests, and environments.

Maps all external-facing endpoints, inputs, and auth boundaries. Scans the whole codebase and produces threat-analysis.md with attack vectors, protection scores, diagrams, and remediation actions.

Reads the full codebase, detects service architecture and dependencies, then generates operational runbooks for common failure modes into runbook.md.

Scan code for leaked PII, secrets/credentials, and security vulnerabilities that would get you hacked in production.

This skill should be used when the user asks to "evaluate a skill", "review skill quality", "score my skill", "check skill best practices", "rate my skills", "evaluate all skills", "compare skills", or wants to assess skill quality across criteria like clarity, token efficiency, anti-cheating, quality gates, determinism, scope discipline, error recovery, observability, and idempotency.

Verified Spec-Driven Development (VSDD) workflow. Fuses SDD + TDD + VDD into a single AI-orchestrated pipeline. Enforces: Spec Crystallization -> Test-First Implementation -> Adversarial Refinement -> Formal Hardening -> Convergence. Use when user invokes /vsdd.

Expert knowledge of Java 25 features including virtual threads, pattern matching, records, sealed classes, value types, and Project Loom/Valhalla features. Use when the user asks about Java programming, Java 25 features, concurrency, performance optimization, or modern Java idioms.

Expert knowledge of Spring Boot 4 and Spring AI 2.0 including auto-configuration, ChatClient, advisors, tool calling, agent patterns, and Spring Framework 7 features. Use when the user asks about Spring Boot, Spring AI, dependency injection, REST APIs, agent configuration, or Spring ecosystem questions.

Full-stack development workflow that orchestrates deployer agents (backend, frontend, database, testing, review) in a phased pipeline. Asks which backend language (Java/Go/Rust) then runs all agents.

Full-stack development workflow that orchestrates deployer agents (backend, frontend, database, testing, review) in a phased pipeline. Asks which backend language (Java/Go/Rust) then runs all agents.