text-humanizer

karanivincent

Remove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound more natural and human-written. Based on Wikipedia's comprehensive "Signs of AI writing" guide. Detects and fixes patterns including: inflated symbolism, promotional language, superficial -ing analyses, vague attributions, em dash overuse, rule of three, AI vocabulary words, negative parallelisms, and excessive conjunctive phrases.

github-image-downloader

karanivincent

Download images attached to GitHub issues for design review workflows. WHEN TO USE: After reading a GitHub issue that contains image attachments (URLs matching https://github.com/user-attachments/assets/...). Do NOT use for issues without images. This skill handles the authenticated download that regular curl/wget cannot perform. TRIGGERS: - Issue body contains github.com/user-attachments/assets/ URLs - User is reviewing designs from a GitHub issue - User asks to see/download images from an issue DO NOT USE when: - Issue has no image attachments - Images are hosted elsewhere (not user-attachments)

skill-creator

karanivincent

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.

pr-reviewer

karanivincent

AI code review using patterns learned from Nicolas and Vincent's 2,768 review comments. Use when reviewing PRs, diffs, or code changes. Triggers on "/pr-reviewer", "review this PR", "review my changes", "code review", or when asked to check code before merging.

pr-slack-announcer

karanivincent

Generate Slack announcement messages for pull requests. Use when user: - Asks to create a Slack message for a PR - Wants to announce a PR in Slack - Needs a PR summary for team communication - Mentions "Slack message", "announce PR", "PR announcement"

claudemd-editor

karanivincent

Create, review, optimize, or troubleshoot CLAUDE.md configuration files. Use when user wants a new CLAUDE.md ("I need a CLAUDE.md for my project"), asks for review/improvements ("review my CLAUDE.md", "make it better"), reports instruction-following issues ("Claude keeps using npm instead of pnpm"), wants to consolidate scattered documentation into CLAUDE.md, or mentions CLAUDE.md structure, best practices, or optimization.

issue-executor

karanivincent

End-to-end issue resolution orchestrator. Takes a Linear issue URL and autonomously drives it from investigation to reviewable PR. Use when: "work on this issue", "execute this issue", "fix this issue", or when pasting a Linear issue URL. For bugs: routes to lightweight (single-agent) or full (agent team) flow based on complexity. Flexible verification by bug type. For stories: interactive brainstorming, then autonomous implementation.

autocompact-threshold

karanivincent

Set the autocompact threshold percentage for context management. Use when adjusting when Claude Code compacts conversation history.

branch-cleanup

karanivincent

Clean up stale git branches (merged or abandoned). Use when user mentions: branch cleanup, delete old/stale/merged branches, prune branches, too many branches, remove unused branches, git housekeeping, cluttered branches, "what branches can I delete", find stale branches, clean up git, branch management, or after merging PRs and wanting to tidy up.

sveltekit-test-guide

karanivincent

Use when writing tests, reviewing test code, or deciding what/how to test in SvelteKit/Svelte projects. Covers unit tests (Vitest), component tests (@testing-library/svelte), and E2E tests (Playwright). Triggers on "write tests for", "add test coverage", "create e2e test", "fix flaky test", "test fails", "test timeout", or any testing-related task.

story-breakdown

karanivincent

Transform user stories into actionable frontend task breakdowns with automatic codebase research. Use when (1) user asks to "break down this story/feature/task", (2) planning implementation for a feature, (3) estimating user stories, (4) preparing task breakdown for sprint planning, or (5) analyzing what's needed to implement a feature in SvelteKit.

pr-creator

karanivincent

Create GitHub pull requests with focused descriptions. Use when user: - Asks to create a PR or pull request - Wants to push changes and open a PR - Needs a PR description written - Mentions "open PR", "create PR", "make a pull request", "submit PR", "can you create a pr"

design-implementer

karanivincent

Orchestrate parallel implementation of a design document using Claude Code agent teams. Reads a design doc, analyzes complexity to determine optimal team structure (single-agent or multi-agent), spawns implementer agents in isolated worktrees with a dedicated reviewer. Uses an umbrella branch pattern: sub-PRs merge to a base branch, then one umbrella PR goes to staging for user review. Reviewers communicate directly with implementers for fix loops — the lead only handles merges and coordination. Fully autonomous — analyzes streams, determines team size, and proceeds without user confirmation. Use when asked to "implement this design", "execute this plan with agents", "build features from design doc", or "run agent team on design".

feature-planner

karanivincent

Orchestrates a comprehensive research and planning workflow for features, bugs, or improvements. Spawns context-gathering agent, then runs parallel planning agents, and consolidates into a final plan for user approval. Use when implementing features, fixing bugs, refactoring code, adding capabilities, or when the user needs help planning complex changes that require understanding the codebase first.

skill-reviewer

karanivincent

Review and analyze skills after real-world usage to identify improvement opportunities. Evaluates runtime effectiveness (workflow gaps, tool issues, edge cases, automation opportunities) and structural quality (token efficiency, progressive disclosure, frontmatter). Produces a scored review report with prioritized, actionable recommendations. Triggers: "review this skill", "review the skill I just used", "what can we improve about this skill", "skill review", "how did that skill perform", "analyze skill", "/skill-reviewer [skill-name]". Use after completing a task that involved a skill, while the session context is fresh.

pr-comment-resolver

karanivincent

Systematically process PR review comments with critical analysis, not blind compliance. Use when: - User says "address PR comments", "resolve review feedback", "fix PR review" - User invokes /pr-comment-resolver with PR number or URL - User needs to respond to code review comments on a pull request - User wants to work through reviewer feedback methodically Supports --auto flag for fully autonomous processing (no prompts, batch analysis, auto-post, auto-resolve threads).

env-sync

karanivincent

Manage environment variables across monorepo workspaces and deployment services. Maintains a central env registry, keeps .env.example files in sync, detects duplicate/aliased variables, and propagates vars to Vercel/Render via MCP tools. Three modes: (1) Add mode - register a new env var, update .env.example files, check deployments. Triggers: "add env var", "new env variable", "I need a new API key", "add SECRET to env". (2) Audit mode - full scan for drift, orphans, missing deployment vars, alias duplicates. Triggers: "env audit", "check env vars", "sync env", "env drift", "are env files up to date". (3) Deploy mode - push missing vars to Vercel/Render. Triggers: "deploy env vars", "sync to vercel", "push env to render", "env deploy". Also handles bootstrapping when no registry exists yet.

test-fixer

karanivincent

Fix failing E2E (Playwright) and unit tests (Vitest), add new E2E tests, and debug test issues using Playwright MCP. Use when (1) tests are failing and need fixing, (2) test selectors or IDs have changed, (3) workflow has changed and tests need updating, (4) adding new E2E tests for features, (5) unit test expectations need updating, (6) debugging why a test fails using live browser interaction.

issue-documenter

karanivincent

Document user stories and bug reports for Linear. Handles two modes: (1) User Story mode - create well-scoped stories with acceptance criteria. Triggers: "write a story", "create a story for", "break down this feature", "scope this work", "new feature". (2) Bug Report mode - investigate and document bugs using Sentry, Vercel, Render, and browser tools. Triggers: "report a bug", "document this bug", "bug report", "something is broken", "log this issue", "this isn't working", "found a bug", "there's an error". Can enrich existing rough Linear issues or create new detailed reports from scratch. This skill ONLY documents - it does NOT suggest fixes, root causes, or solutions.