diegopetrucci

iOS Simulator automation and interaction using the AXe CLI tool. Use when working with iOS simulators for UI automation, testing, accessibility verification, or screen recording. Specific use cases include simulating touches and gestures (tap, swipe), entering text, pressing hardware buttons (home, lock, Siri), recording or streaming simulator video, extracting UI accessibility information, or automating iOS simulator interactions. Assumes axe CLI is already installed.

macOS UI automation and screenshot capture tool. Use when working with macOS applications and needing to: (1) Capture screenshots with UI element annotations, (2) Automate UI interactions (clicking, typing, scrolling), (3) Control applications and windows programmatically, (4) Navigate menus and dialogs, (5) Perform visual analysis of screens, (6) Create multi-step automation workflows. Requires macOS 15.0+ with Screen Recording and Accessibility permissions.

Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.

Sync dotfiles from private chezmoi repository to public dotfiles repository. Use when the user requests to update their public dotfiles repo, sync dotfiles, or copy dotfiles from chezmoi to the public repository. This skill renders chezmoi templates and applies safety filtering to prevent accidentally exposing sensitive files.

Use when installing, updating, auditing, or uninstalling The Last Harness isolated Pi package or one-line installer. Covers the tlh wrapper, isolated Pi profile, settings merges, and rollback documentation.

Use when finishing repository development work for The Last Harness before handoff or review. Covers repo-only final hygiene checks and is not for packaged end-user use.

Sentry CLI tool for error tracking and monitoring workflows. Use when working with Sentry for (1) Creating and managing releases, (2) Uploading debug symbols (dSYMs) for iOS/mobile crash reporting, (3) Uploading source maps for JavaScript/web applications, (4) Associating commits with releases for issue tracking, (5) Managing deploys across environments, (6) Validating debug information files, or any other Sentry-related development and deployment tasks. Assumes sentry-cli is installed via brew and user is authenticated.

How to run swift unit tests

Generate clean, absurd article illustrations in Ian's Xiaohei style. Use this skill when the user wants inline illustrations, shot lists, image edits, or visual metaphors for articles, blog posts, Notion pages, workflows, methods, structures, states, or key ideas. Default all response text and handwritten image labels to English unless the user explicitly requests another language.

Audit the current repository for agent workflow efficiency. Use when the user wants Codex to try setup, build, test, lint, run, or other obvious project commands and report where the agent instructions, AGENTS.md, README, or scripts are wasting context tokens, causing avoidable command exploration, or leaving important steps implicit. This skill optimizes the workflow and instructions for agents; it should not fix product code unless explicitly asked.

Critically evaluate PR review comments against the actual code on the current branch. Use when the user triggers this skill on a branch and pastes PR review comments one at a time. Investigates each comment for validity rather than accepting it at face value, explains the verdict, and if valid asks before implementing the fix. Trigger phrases include "triage PR comments", "review comments", or when the user invokes this skill while on a feature branch.

GitHub research scout. Locates exact code locations across GitHub repos using `gh` (search code, tree, contents API) in an isolated workspace, then returns concise, path-first findings with line-ranged evidence. Use when the answer likely lives in GitHub repos, exact repo/path is unknown, or when you'd otherwise burn turns on exploratory `gh search` / tree probes plus `rg`/`read` on fetched files. Pass scope hints (repo names, owners, refs, paths) in the query when you have them.

Generate clean, absurd article illustrations in Ian's Xiaohei style. Use this skill when the user wants inline illustrations, shot lists, image edits, or visual metaphors for articles, blog posts, Notion pages, workflows, methods, structures, states, or key ideas. Default all response text and handwritten image labels to English unless the user explicitly requests another language.

Plan TfL journeys from start/end/time, resolve locations (prefer postcodes), and warn about disruptions; suggest alternatives when disrupted.

Review a codebase and say what should change if it were being started again from scratch. Use when the user wants a reset-style retrospective, architecture do-over, or "what would you change if you restarted this project?" analysis. Start by asking whether to analyze a single commit, the branch against its base, or the whole project.

Sentry CLI tool for iOS error tracking and monitoring workflows. Use when working with Sentry for (1) Listing and investigating issues and crashes, (2) Uploading debug symbols (dSYMs) for crash report symbolication, (3) Creating and managing releases, (4) Associating commits with releases for issue tracking, (5) Managing deploys across environments, (6) Viewing logs and monitoring cron jobs, (7) Sending test events, or any other Sentry-related iOS development and deployment tasks. Also use when the user mentions "sentry", "crashes", "dSYMs", "symbolication", "debug files", or "crash reports".

Scan the current branch's diff and remove AI-generated code slop — unnecessary comments, gratuitous defensive checks, `any` casts, and style inconsistencies with the surrounding code. Use this skill whenever the user wants to clean up AI-generated code, mentions "slop", asks to make AI code look more human-written, or wants to polish a branch before review. Also trigger when the user says things like "clean up this branch", "remove the AI smell", or "make this code less obviously AI-written".