brianlovin/agent-config
18 skills2k installs6.4k starsGitHub
Install
npx skills add https://github.com/brianlovin/agent-configSkills in this repo
1Knipknip is a ship-phase cleanup skill from brianlovin/agent-config that instructs a coding agent to run Knip against JavaScript or TypeScript codebases. The workflow checks for a local Knip install via `npx knip --version`, adds knip to devDependencies when missing (npm, pnpm, yarn, or bun per lockfile), then reports unused files, package dependencies, exports, and types across the repo. Knip deliberately does not remove unused imports or variables inside files—that remains a linter responsibility. Developers reach for knip when bundles feel bloated, dependency trees are stale, or reviewers flag dead exports before a release tag.635installs2Simplifysimplify is a code refinement skill from brianlovin/agent-config that acts as an expert code simplification specialist for recently modified files. The skill analyzes new or changed code and applies project-specific best practices to improve clarity, consistency, and maintainability without altering behavior. Developers reach for simplify immediately after writing code when diffs need cleanup before review or merge—prioritizing readable, explicit code over overly compact solutions. It focuses on recently modified code only, making it a lightweight post-implementation polish step rather than a full codebase refactor tool.606installs3DeslopDeslop is an agent-config skill that compares the current feature branch to main and removes AI-generated code slop introduced in that branch. It targets extra comments a human would not write, defensive try/catch blocks abnormal for trusted call paths, casts to any that bypass type errors, and style inconsistent with the rest of each file. After cleanup, Deslop reports a 1–3 sentence summary of what changed. Developers reach for Deslop after agent-assisted coding sessions and before opening or updating a pull request, when diffs look verbose or stylistically off compared to the existing codebase.552installs4Ramsrams runs an accessibility and visual design review on UI component code. It checks components against WCAG 2.1 across critical, serious, and moderate severity tiers, then reviews layout, typography, color, contrast, and component states. It reports findings with line numbers, code snippets, WCAG references, and fixes, ending with a 0-100 score, and can fix issues on request.21installs5Agent Browseragent-browser is a command-line browser automation tool that drives Chrome or Chromium directly over the Chrome DevTools Protocol. Developers use it to navigate pages, fill forms, click buttons, take screenshots, extract data, and test web apps from an agent or terminal. It follows a navigate, snapshot, interact, re-snapshot loop and keeps the browser alive between commands via a background daemon.20installs6TddTDD guides test-driven development using a red-green-refactor loop. It tells the agent to write one behavior test, make it pass with minimal code, then repeat, rather than writing all tests up front. It favors integration-style tests that exercise public interfaces so they survive refactors. A developer uses it when building a feature or fixing a bug test-first.17installs7Bunbun is a guidance skill that tells the agent to use Bun as the default JavaScript and TypeScript runtime and package manager. It provides command mappings from Node, npm, pnpm, and Vite to Bun, plus Bun-specific APIs to prefer over Node equivalents. Developers use it so generated code and commands stay Bun-native across the runtime, testing, and frontend dev.16installs8React Doctorreact-doctor scans a React codebase for security, performance, correctness, and architecture issues and outputs a 0-100 score with actionable diagnostics. It runs via npx react-doctor and reports file paths and line numbers so developers can fix errors by severity and re-run to verify. Its 47+ rules span state and effects, bundle size, Next.js patterns, server auth, accessibility, and dead code.16installs9Skill CreatorSkill-creator is a guide for writing new Claude skills or updating existing ones. It explains the required SKILL.md structure with YAML frontmatter and the optional scripts, references, and assets folders. A developer uses it when packaging specialized knowledge or workflows so an agent can trigger and follow them. It emphasizes concise, token-efficient authoring and matching instruction specificity to task fragility.15installs10Find Skillsfind-skills helps the agent discover and install skills from the open agent skills ecosystem. When a user asks how to do something or wants a skill for a task, it searches with the Skills CLI (npx skills find), presents matching skills with install commands and links, and can install them globally. It is meant to extend an agent's capabilities on demand.14installs11Grill Megrill-me is a short skill that has the agent interview the user relentlessly about a plan or design. It walks down each branch of the decision tree, resolving dependencies between decisions one at a time and offering a recommended answer per question. When a question can be answered from the codebase, it explores the code instead of asking.13installs12Reclaudereclaude refactors CLAUDE.md files to follow progressive-disclosure principles. It reports line count against thresholds, integrates workflow orchestration, ensures a Verification section exists, finds contradictions, and identifies content to extract into global skills or move into .claude/rules/. It targets a concise root CLAUDE.md with links to detailed rules files.13installs13Chrome Webstore Release Blueprintchrome-webstore-release-blueprint is a hands-on setup guide for automating Chrome extension releases through the Chrome Web Store API. It leads a user through enabling the API, configuring OAuth, generating a refresh token, capturing store IDs, and wiring local and CI secrets. It then designs a version-triggered publish workflow that builds, uploads, and checks submission status.12installs14Faviconfavicon is a skill that generates a complete favicon set from a single source image and wires it into a web project. It verifies ImageMagick, detects the framework to find the static assets directory, and produces favicon.ico, PNG icons, apple-touch-icon, and manifest icons. It then creates or updates site.webmanifest and adds the matching favicon link tags to the HTML or layout files.12installs15Fix Sentry Issuesfix-sentry-issues is a workflow for discovering, triaging, and fixing production errors using the Sentry MCP server. It treats each error as a signal, builds an Investigate-or-Ignore triage table, then investigates one issue at a time by pulling event data, reading the failing code path, tracing inputs upstream, and reproducing. It fixes the root cause on a per-issue branch and resolves the Sentry issue only after merge.11installs16Playwriterplaywriter is a skill that controls the user's already-open Chrome tab through the Playwriter CLI instead of launching a new browser. Developers use it to inspect live UI state, run scripted browser actions, capture console output, and reproduce frontend issues directly in the user's tab. It attaches to a session and runs Playwright-style page commands against that tab.10installs17Sentrysentry is a guidance skill describing how to instrument a Next.js application with Sentry for error monitoring and performance tracing. It shows capturing exceptions with Sentry.captureException, creating spans with Sentry.startSpan for UI actions and API calls, configuring the client, server, and edge init files, and using structured logging. Developers use it to add consistent Sentry instrumentation to their code.10installs18WorkflowWorkflow orchestrates complex coding tasks by enforcing structured execution. It defaults to plan mode for any non-trivial task, uses subagents to keep context clean, and requires verification before marking work done. It also runs a self-improvement loop that records corrections in tasks/lessons.md. A developer uses it to impose discipline on multi-step implementations, refactors, and bug fixes.10installs