Bun plugins

Provide expert guidance for building, debugging, and optimizing TypeScript and JavaScript applications with modern frameworks like Next.js and Node.js, covering ES6+ patterns, React performance, App Router, framework selection, and advanced type-level programming.

Migrate Lodash code to es-toolkit in JavaScript and TypeScript projects by replacing imports and comparing APIs to shrink bundle sizes, get function recommendations matching your needs or code with imports examples and docs, and follow tailored setup guides for Node.js Bun Deno and browsers to optimize performance.

Blocks destructive git and filesystem commands before execution by intercepting shell commands and validating them against configurable safety rules at user, project, or GitHub scope.

Automate the entire PRP workflow — plan, implement, debug, review, commit, and create PRs — using specialized agents that investigate issues, generate implementation plans, run validation loops, and execute multi-aspect code reviews.

Schedule and run Claude prompts automatically with a cron-like daemon that manages heartbeats, cron jobs, and integrates with Telegram and Discord for notifications and control.

Receive phone calls from Claude when it needs input, wants to report progress, or needs to discuss a task — enabling voice-based interaction when text is insufficient.

Read, write, search, and surgically edit Obsidian vault notes, tags, and frontmatter via the Model Context Protocol over STDIO or Streamable HTTP.

Manage an AI-native job queue for Bun with SQLite persistence, including queue setup, workers, cron scheduling, retries, dead-letter queues, and debugging via a specialized agent and an interactive queue manager.

Generates business intelligence dashboards, infographics, AI images/videos, and automates posting to social media (WeChat, X, Xiaohongshu) with tools for document parsing, Obsidian integration, and browser automation.

Enables AI agents to interact with over a dozen messaging platforms via CLI, automating message send/receive, channel management, reactions, and authentication with automatic credential extraction.

Bridge WeChat messaging to Claude Code — receive and respond to AI queries from WeChat contacts via an iLink bot with built-in access control (allowlists, pairing requests, DM policy).

Generate AI images from text prompts using a fast CLI powered by Gemini 1.5 Flash or Pro models. Set resolutions from 512px to 4K, custom aspect ratios, exact dimensions, style reference images, green screen transparency for compositing, and track costs per generation directly in terminal.

Run and confirm outputs of verification commands like tests, builds, lints, and type-checks before claiming task completion, fixes, or readiness for commits/PRs.

Automate headless browser testing with Bun using persistent WebView sessions. Perform UI interactions like navigation, clicking, form filling, screenshots, and content verification via CLI.

Trace bugs backward through call stacks to identify original triggers for errors deep in execution, long stack traces, unclear data origins, and test pollution.

Work with 300+ AI models through OpenRouter: query models by price/capabilities, run analytics on usage data, generate images/video/speech, transcribe audio, and integrate LLM calls into TypeScript agents with CLI tools or OAuth authentication—all without managing individual API providers.

Set up, configure, extend, and debug Better Auth TypeScript authentication for Next.js, Nuxt, Cloudflare D1 via Drizzle/Prisma/Kysely/Hono, and 15+ frameworks. Interactively generate auth configs, schemas, routes, clients, and env vars; add plugins like 2FA, passkeys, Stripe; explain errors with fixes; reference providers; autonomously diagnose issues.

Set up Cloudflare Queues via interactive wrangler wizard for producers, consumers, batching, retries, and dead letter queues; monitor real-time metrics like backlogs and DLQ status; troubleshoot errors such as timeouts, retries, and throughput issues; optimize performance with concurrency and scaling recommendations using commands and diagnostic agents.

Securely upgrade dependencies in JavaScript/TypeScript projects across npm, Yarn, pnpm, Bun, and Deno with supply chain protections including cooldowns, script hardening, lockfile validation, compatibility analysis, staged rollouts, and CI integration via Dependabot/Renovate.

Set up Nuxt UI v4 with Tailwind v4 in existing Nuxt 4 projects, migrate from v2/v3, scaffold typed Vue components for dashboards/forms/modals/chats, generate semantic themes, select optimal components for layouts/overlays, access remote docs, and diagnose/fix styling/composable/TS errors.

Set up Cloudflare R2 buckets in Workers projects via wrangler, bind them to your code with TypeScript types, upload and download objects including large files via multipart with Hono, generate presigned URLs, debug and fix CORS errors, configure event notifications for automation, and migrate S3 buckets using rclone.

Build stateful real-time apps like chat rooms and multiplayer games using Cloudflare Durable Objects: scaffold projects with wrangler configs, TypeScript boilerplate, and tests; generate patterns for WebSockets, alarms, and SQLite; interactively debug issues, optimize performance bottlenecks, and manage class migrations with validation.

Run mutation testing with Stryker on TypeScript/JavaScript projects using Vitest or bun test, and mutmut on Python projects to identify weak unit tests that pass despite code mutations, improving test suite quality and effectiveness.

Discover, analyze, and execute tools, prompts, and resources on MCP servers via CLI scripts. Integrate capabilities, filter tools, manage configurations, resolve context bloat, and handle execution errors programmatically for AI workflows.

Build durable long-running multi-step workflows on Cloudflare Workers with automatic retries, state persistence, sleeps, and event handling. Scaffold TypeScript classes, setup projects interactively, test locally/remotely, debug failures like NonRetryableError or timeouts, and optimize code for cost, speed, and reliability using commands and agents.

Manage Cloudflare Workers KV end-to-end: interactively set up production/preview namespaces with wrangler bindings and TypeScript types, test CRUD operations and connections, analyze/optimize code for performance/cost via anti-pattern detection and refactoring agents, debug KV_ERROR/429 limits/consistency issues with diagnostics and fixes.

Analyze test suites in TypeScript, JavaScript, Python, Playwright, and Bun to detect smells like overmocking, flaky tests, fragile tests, poor assertions, and coverage issues. Assess test correctness, reliability, maintainability, and overall quality during reviews or refactoring workflows.

Generate, prototype, and deploy full-stack web apps from natural language descriptions or design mockups, with optional SaaS features like Stripe billing and auth, all deployed to Cloudflare Workers without a local build step.

Orchestrate parallel subagent workflows with ultrawork mode, automated git/PR management, TDD enforcement, Ralph Loop autonomous execution, and structured planning with PRD generation, all within Claude sessions

Connect WeChat accounts to Claude Code via QR code pairing and official iLink Bot API, manage user access with allowlists and policies (pairing/allowlist/disabled), and run a local Bun-based MCP server for WeChat messaging, bot tools, and AI coding assistant integration.

Search PubMed and Europe PMC biomedical literature, fetch full-text articles via PMC/EPMC/Unpaywall, retrieve citations and MeSH terms, and integrate the results into AI agent workflows through the MCP protocol.

Interact with Google NotebookLM via CLI to generate artifacts like slides, audio overviews, videos, mind maps, quizzes, flashcards, infographics, reports, and data tables from notebooks or uploaded sources. Chat with AI on sources, manage notebooks, notes, research, and sources.

Integrate You.com's web search, research with citations, and content extraction into AI agents built with frameworks like Vercel AI SDK, Claude Agent SDK, OpenAI Agents SDK, crewAI, LangChain, and Microsoft Teams.ai. Also accessible via direct REST API or bash CLI.

Coordinate multi-agent coding workflows by bridging Claude Code with a shared daemon for push channel delivery, bidirectional replies, and turn-based collaboration between agents.

Run TypeScript scripts on Claude Code session events and enable Claude to browse and retrieve files from a local meme directory via an MCP server.

Search and retrieve clinical trial data from ClinicalTrials.gov, including study details, locations, and outcomes, with MCP-based integration for programmatic access.

Compile human experts into installable Claude Code plugins acting as AI agents by conducting structured behavioral interviews, analyzing work artifacts with MCP tools, and generating personalized digital twins that replicate decision-making, communication, and expertise.

Set up Claude Ultimate HUD status line to display rate limits, project info, tool activity, todo progress, with improved emoji/CJK support, detecting platform and configuring for language/plan. Automate releases via /release: build with bun, bump versions across files, generate bilingual changelogs from git diff, update READMEs, commit/push, create GitHub release.

Enqueue PRD files or descriptions for AI-orchestrated builds via eforge daemon, monitor run status and queue, configure models and agents interactively, initialize projects with package detection, update the CLI, and restart daemon safely—all from Claude Code conversations.

Invoke Google Gemini CLI within Claude Code workflows to analyze full codebases using 1M+ token context for second opinions, architectural advice, code reviews, security audits, and debugging.

Debug and inspect React Native/Expo apps by connecting to the Metro bundler via the Chrome DevTools Protocol, enabling runtime inspection, module resolution, and JavaScript bundling automation from the command line.

Chat with Claude through WeChat (微信) with support for text, images, files, voice, video, and streaming replies. Manage WeChat and Feishu channel access control including pairing approvals and allowlists.

Enforce runtime security policies for AI coding sessions: block risky operations, audit every action, scan for vulnerabilities, and hunt threats—with real-time dashboards and automated policy checks.

Build autonomous agents and multi-step workflows with Anthropic's Claude using a TypeScript SDK. Orchestrate subagents and tools, manage MCP servers and sessions, handle permissions, and resolve errors like CLI not found or context exceeded.

Streamline TypeScript development workflows using Bun for dependency management, building executables/bundles, running tests, debugging via inspector, and publishing to npm. Lint/format with Biome, eliminate dead code using Knip, and add Sentry monitoring for errors and performance.

Manage project tool versions, environment variables, and build tasks with Mise. Install and pin versions for Node, Python, Rust, Go, and more; define tasks with dependencies, arguments, and env vars; and enforce configuration consistency.

Manage Bun.js project lifecycle: bundle JavaScript/TypeScript for multiple targets, manage dependencies with Bun package manager, write and run tests with Bun's test runner, and interact with SQLite databases. Includes automation hooks for cleaning up orphaned worktrees.

Set up isolated git worktrees for feature branches with automatic dependency installs, port allocation, and baseline tests. Generate bash scripts to manage dev server lifecycles—start, stop, status, ports—using pool-based isolation, detected from project structure and package managers like pnpm, bun, yarn, npm, Cargo, or Go.

Build production-ready backend AI in Node.js/TypeScript/Bun apps using Vercel AI SDK v5 for text generation, structured outputs, tool calls, and agents across OpenAI, Anthropic, and Google providers. Resolves AI_APICallError, AI_NoObjectGeneratedError, and streaming issues for reliable server-side AI workflows.

Integrate Google Gemini embeddings API (gemini-embedding-001) into TypeScript apps for RAG pipelines, semantic search, and Cloudflare Vectorize vector stores. Implement batching, task types via TypeScript SDK, and resolve errors like dimension mismatches, rate limits, text truncation.

Build and manage stateful OpenAI Assistant chatbots using API v2 with persistent threads and runs, enabling Code Interpreter for code execution, File Search over uploads, RAG via vector stores. Poll for completion, resolve active run errors and indexing delays, and migrate to Responses API. (Note: sunsets Aug 2026.)

Deploy and manage full Cloudflare account services—Workers, KV storage, R2 buckets, Pages projects, DNS records, and routes—via API calls. Automatically validate credentials, handle deployments, and extract service URLs for immediate use in Bun runtime workflows.

Add zero-config animations to React lists, accordions, and toasts with @formkit/auto-animate, preventing SSR errors, conditional rendering issues, and missing keys in Next.js, Vite, Bun, and Cloudflare apps.

Set up production-ready shadcn-vue UI components in Vue/Nuxt apps with Tailwind CSS and Reka UI, enabling accessible forms, data tables, charts, dark mode, and easy component imports.

Build and deploy Puppeteer/Playwright browser automation workflows in Cloudflare Workers to generate screenshots and PDFs, scrape websites, manage sessions, and handle rendering errors, timeouts, and memory limits at the edge.

Perform RAG-powered search and Q&A on 100+ file formats like PDFs, code, and Markdown using Google Gemini. Build document Q&A systems and knowledge bases while handling immutability errors, quotas, and polling failures with Gemini 1.5 Pro/Flash support.

Configure Cloudflare Email Routing to receive incoming emails at the edge, parse them with postal-mime in Workers, forward to verified addresses, deploy via wrangler, and resolve SPF/DKIM issues for seamless email workflows.

Run Puppeteer CLI scripts to automate browser tasks from your terminal, capturing screenshots, analyzing performance, monitoring network requests, scraping web data, filling forms, evaluating JavaScript, and debugging automation errors.

Integrate OpenAI APIs into Node.js apps using SDK examples for chat completions (GPT-4o, GPT-5.2), embeddings, image generation, Whisper transcription, TTS audio, streaming responses, function calling, and error handling to build AI-powered features quickly.

Integrate Google Gemini API into TypeScript, JavaScript, or Bun projects with the @google/genai SDK to enable multimodal generation, function calling, thinking mode, and streaming. Migrate seamlessly from the deprecated @google/generative-ai package while fixing context length and multimodal format errors.

Enables natural language messaging between local Claude Code sessions for multi-agent coordination and communication.

Bridge NATS messaging into Claude Code sessions: configure channels and permissions via CLI, then publish-subscribe, request-reply, and queue messages directly using a local Bun runtime—no secrets required.

Gives Claude Code persistent memory across sessions by automatically extracting facts, actions, and decisions from conversations, running nightly reflections on user patterns and knowledge gaps, and performing end-of-day memory maintenance to prune stale information and update state files.

Persist Claude Code session context across conversations using Honcho's knowledge graph, injecting relevant memories on user prompts and configuring memory preferences via interactive menus and first-time setup guidance.

Orchestrate spec-driven development and product workflows using MUX multi-agent delegation. Trigger /spec, mux ospec, or mux-roadmap to automate stages like research, planning, implementation, testing, reviews, documentation, and git commits with user confirmations and cross-session continuity.

Transform academic papers and research queries into executable code with tests, educational comics, visual schemas, structured reports with citations, and AI-generated images. Process PDFs, arXiv URLs, or prompts via specialized skills for rapid implementation, visualization, and deep analysis workflows.

Generate publication-ready technical articles and social media posts in your author voice, with web research, verified runnable companion code projects (React, Laravel, Node, Python), and multi-language translations. Queue tasks in SQLite for interactive single-article creation or autonomous batch processing.

SDD for Claude Code walks a feature through the full lifecycle — from idea capture through spec, architecture, data model, API design, task breakdown, TDD implementation, review, and shipping — using gated atomic skills and specialized subagents. It maintains an outcome-oriented roadmap, a living glossary, ADRs, and test plans, and surfaces a dashboard UI for pipeline visibility.

Control Android and iOS devices using natural language — inspect UI state, run tests, and execute device interactions via an MCP server enhanced with PR/CI workflows and dead-code removal.

Generate complete slide decks in PPTX or PDF from raw content via RDIV workflow: analyze input, interactively select styles, auto-generate outlines and prompts, batch-produce images with PaperBanana plugin, then merge into polished presentations.

Automate Office document workflows in Claude Code by inspecting PPTX/DOCX/XLSX contents, building files from JSON specs, extracting text/data, verifying results via CLI, and directly editing PowerPoint slides, layouts, images, shapes, and templates through XML manipulation.

Integrate Bankr x402 SDK into TypeScript/Node.js projects to query multi-chain token balances/NFTs, execute swaps/transfers/bridges/staking, manage async jobs/wallets via natural language prompts. Scaffold bot/web/CLI/dashboard templates with AI-assisted Web3/DeFi workflows on Base/Eth/Polygon/Solana.

Scrape local events from Instagram accounts, Facebook pages/groups, and web aggregators into a SQLite database. Update event details via re-scraping and generate dated Markdown newsletters customized by natural language formatting preferences from config sources.

Convert PDFs, EPUBs, HTML, images, YouTube videos, audio files, URLs, and other media to clean Markdown using a local MCP server. Extract and process content directly in your IDE or workflow without uploading to external services.

Manage and automate YAML workflow playbooks for LLM agents via MCP, including storing, querying, and creating playbooks with built-in scaffolding, testing, deployment, and security auditing for MCP servers built on @cyanheads/mcp-ts-core.

Interact with Claude Code via Telegram bot: receive notifications for task updates, blockers, clarifications, permission requests, and idle prompts; respond directly in chat to unblock workflows without checking the terminal. Setup requires bot token and chat ID; uses local MCP server and hooks for event triggers.

Enforce strict Red-Green-Refactor cycles with recorded proof of test failure before implementation, while using Locked Intent Boundaries to prevent coding agents from redefining goals during planning.

Subagent dispatch, browser automation, and prompt engineering within Claude Code. Perform web research, screenshot analysis, and adversarial prompt review. Humanize prose and decompose ambiguous requests.

Deploy websites, apps, and AI agents to Agentuity's cloud platform with managed databases, storage, queues, cron jobs, and real-time frontend hooks

Manipulate Office documents (PPTX, DOCX, XLSX) by inspecting contents, previewing visuals, extracting data, building from JSON specs, and verifying integrity using aurochs CLI on Bun.

Enables peer-to-peer direct messaging between active Claude Code sessions on the same machine via a shared SQLite bus, allowing session registration, direct messaging, and broadcast for multi-session coordination.

Orchestrate autonomous development workflows with a team of specialized AI agents for coding, testing, security, DevOps, and documentation, plus 70+ skills for tasks like CLI demos, git visualization, UI design, and SaaS launch audits.

Display, customize, and interact with ASCII art companion pets in your Claude Code statusline. Generate sprites with species, eyes, hats, rarity, and shininess; manage pet collections; trigger petting animations and hearts; toggle screensaver modes with random or collection intervals; view pet stats, speech bubbles (fun or review), and interaction history logs.

Bulk scan, transcribe, and summarize videos from YouTube channels, podcasts from RSS feeds like Apple Podcasts and Spotify, and SoundCloud content using yt-dlp. Generate consolidated Markdown research notes for media analysis workflows.

Develop Bun projects with accurate knowledge of versions 1.2 through 1.3.10 features, including route-based HTTP servers, built-in clients for Redis, PostgreSQL, MySQL, SQLite, S3 storage integration, test runner, bundler, and package manager updates.

Run Playwright MCP server in Claude Code to automate browser tasks via accessibility tree: navigate pages, click elements, fill forms, capture screenshots, and extract content from blocked sites or local UIs for testing and scraping workflows.

Automate your Obsidian second-brain vault with AI-powered skills for knowledge extraction, code generation, issue tracking, email management, and deep thinking. Runs autonomous workflows for vault maintenance, book insights, TDD development, PR merges, and principle-based audits via multi-agent teams.

Persists project goals, decisions, gotchas, and patterns across sessions using a SQLite-backed long-term memory system. Recalls relevant context automatically, mines insights from git history, and provides an interactive onboarding wizard to seed project knowledge.

Batch review GitHub PRs and GitLab MRs via live tmux dashboard spawning isolated sessions per PR, delegate peer reviews gathering context for errors/types/tests with style-matched comments, self-review code changes using visual diff viewer to add line-specific comments for git edits, and automate follow-ups classifying threads with status summaries and approve suggestions.

Generate and look up SeeFlow flows from natural-language prompts, creating validated flow.json and node-sidecar files that serve as architectural ground truth for your project.

Capture one-line insights or conversation knowledge into a typed, tagged sp-context base with auto-inferred titles and related doc searches; sync high-value claude-mem observations like decisions and bugs; run sp doctor for automated health checks fixing tags, links, metadata, staleness, indexing, and git commits to maintain a clean company knowledge repository.

Persistent project memory for AI coding agents: reindexes your codebase on file changes, surfaces context via semantic search and dependency analysis, traces bugs to root cause, plans changes with blast-radius assessment, and resumes work across sessions by recording decisions and next steps.

Replace Claude Code's context compaction with Morph's service by installing an API key in ~/.claude/morph/.env and appending override instructions to ~/.claude/CLAUDE.md. Uninstall cleans up files. Run TypeScript hooks via Bun on PreCompact, SessionStart, SubagentStart, and UserPromptSubmit for session lifecycle automation.

Launch a pixel art viewer at localhost:7777 simulating a Gather Town-style IT startup office where AI agents generate 2-4 turn Korean dialogues reflecting team personalities in scenarios like bugs, deployments, and lunches. Start/stop/toggle the server with bun/bash commands and run custom bash scripts on 8 Claude Code lifecycle events.

Manage Things 3 task manager from Claude: capture todos, read and query tasks, triage the Today list, and create/update projects — all via CLI and URL schemes on macOS.

Launch and manage Dwarf Fortress-inspired ASCII simulations in tmux sessions overseen by Claude: strike the earth to generate dramatic fortress names and embark on new worlds, or return to the depths to resume saves via interactive picker—all via CLI commands powered by a local Bun/TypeScript MCP server.