Jewgah
UserBattle-tested Claude Code skills: deep multi-agent code review (/review-deep), adversarial plan review (/review-plan), automation scout-and-builder (/loopable), agent-team scaffolder (/init-agents)
Categories
Indexed Skills (12)
battle-test
Run the frontend and backend checks that CAN actually be run for the current change, then report what ran, a battle-tested score out of 100, and the checks that still need a human. Use after implementing a feature or a bugfix and before committing it: "test this", "run the tests", "did I break anything", "how solid is this change", "what still needs manual testing", "teste ça", "est-ce que c'est bien testé". Runs the repo's own tooling only, never a new framework. It reports; it never commits, pushes, or deploys.
init-agents
Scaffold a tailored Claude Code agent team into the current repo — detects the stack, proposes agents/commands/hooks, then writes and wires them into .claude/ and CLAUDE.md.
loopable
Find recurring work worth automating, then take the top candidate through the full delivery flow — plan, /review-plan, implement, /review-deep, fix all findings, commit, and file a ticket in ~/Downloads/tickets/{project}/.
loopit
Drive a task queue (loop-tasks.md) to completion in ANY repo. Detect the repo's own rules, then for each task plan -> /review-plan -> implement -> verify -> /review -> fix all findings -> /battle-test -> commit -> push -> client doc -> mark done. Each task runs in its OWN subagent (context isolation, token-cheap); the main session only orchestrates. Autonomous between tasks; pauses only for genuine product decisions.
review-deep
Deep multi-agent code review — parallel agents cross-check each other to catch critical bugs
review
Review code changes for flaws, bugs, security issues, and potential problems
vulgarize
Turn the technical changes in the working tree into a short, casual, jargon-free message you can send straight to the client (en/fr)
security-audit
Audit the codebase for common security vulnerabilities (OWASP, secrets, auth, injection)
choices
Summarize the open decisions/choices from the current conversation into a simple, copy-paste WhatsApp message for the team to weigh in on. Use when work is blocked on product/scope decisions a human must make.
commit
Commit staged changes, update CLAUDE.md and memory if relevant
frontend-verify
Verify frontend changes end to end after editing a web app, instead of manually clicking through pages. Use this whenever you have changed UI code and need to confirm nothing broke: "verify my frontend", "check the site after these edits", "did my UI break", "did my changes break anything", "make sure these routes still work", "smoke test the app", "check for console errors", "validate the pages I touched". Works with Next.js (app and pages router), React, Vite, and any local dev server. Built to be token cheap: it reads console errors and failed network requests first and writes full page state to disk, so it only pulls a snapshot or a screenshot into context when a route actually fails. Use it before saying a frontend change is done.
sessions
List recent Claude Code sessions across all projects with project name, session ID, resume command, and short description of what the session was about. Handles multiple parallel sessions per project. Use when the user wants to find, resume, or recall a previous Claude conversation — especially before/after a reboot.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.