npm plugins

Author, edit, validate, and test Microsoft Copilot Studio agents using YAML files. Clone agents from the cloud, modify topics, actions, knowledge sources, and global variables, then push changes back. Run evaluations and batch tests against draft or published agents.

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.

Automate software development workflows on GitHub using AI agents: from issue triage and feature implementation to code review, deployment validation, and incident response, all within an orchestrated pipeline that treats GitHub issues as structured knowledge graphs.

Manage Nx monorepos through Claude: run generators, execute tasks, explore workspace structure, fix dependency resolution errors, and drive autonomous CI fix loops that retrieve failures, analyze errors, apply fixes, and iterate until green.

Orchestrate the full software development lifecycle from idea to shipped code — planning, TDD implementation, code quality audits, architectural refactoring, testing, deployment readiness checks, and structured git workflows with conventional commits and npm publishing.

Orchestrate multi-specialist architectural reviews using AI agents for security, performance, domain-driven design, and maintainability, create and track Architecture Decision Records (ADRs), enforce YAGNI principles with pragmatic guard mode, and auto-install the framework into new projects based on tech stack analysis.

Manage Skyboard kanban boards on AT Protocol from terminal CLI: join boards by URI, list boards/columns/cards, create/edit/move/delete/comment cards, view details, login/logout with OAuth.

Generate AI images and videos from text prompts via Leonardo.Ai CLI in your terminal. Specify models, dimensions, enhancements; poll job status, download results to files/directories; list models and check credits.

Run a complete PDCA (Plan-Do-Check-Act) knowledge work OS inside Claude Code: decompose tasks into parallel worktrees, research blocked URLs via adaptive fetch chains, review drafts with multi-perspective agents, and evolve prompts through benchmark-driven optimization loops.

Automatically detect repeated multi-step workflow patterns in Claude Code sessions, generate reusable skills, and load them dynamically mid-session to automate common tasks like search-and-fix or find-and-update.

Identify emerging developer trends before they hit mainstream by scanning GitHub trending repos, package registry download spikes (PyPI, NPM), and tech Twitter conversations.

Generate full architectural analysis reports mapping project components, dependencies, relationships, risks, bottlenecks, and infrastructure; audit dependencies across languages and package managers for vulnerabilities, outdated libraries, CVEs, and licenses; deeply analyze individual components for internal structure, business rules, design patterns, coupling, security, and technical debt. Produces timestamped Markdown reports saved to /docs/.

Gives Claude UI design superpowers: build, refactor, and polish web UIs with structured MCP workflows for typography, layout, motion, and accessibility. Also converts raster logos to SVG and manages npm releases.

Build Pi coding agents with grounded skills: deploy vLLM GPU pods on DataCrunch/RunPod, integrate via RPC/Node SDK in TypeScript apps, embed Lit web UIs and TUI components, author extensions/packages, customize prompts/behaviors, manage CLI workspaces with tmux, and deploy Dockerized Slack bots.

Launch and manage spec-driven marketing campaigns for developer products with automated positioning enforcement, quality gates, SEO audits, brand-voice content generation, and Vercel landing page deployment. Handles the full campaign lifecycle from brief to archive, including drift detection and learning extraction.

Build and customize AI chat interfaces in React/Next.js apps using assistant-ui library. Scaffold projects via CLI with templates, customize UI primitives and state with hooks, add multi-thread management and cloud persistence with auth, integrate LLM tools and streaming backends, and migrate packages to latest versions.

Automate Indian stock market research workflows: fetch concall transcripts from screener.in/BSE/NSE, extract variant perception scorecards and growth triggers via AI analysis, generate equity deep-dive reports with NotebookLM, and publish professional PDFs/HTML to Netlify.

Manage Business Central AL development with an engineering manager agent that spawns parallel specialist teams for requirements gathering, competitive solution design, multi-developer implementation, 4-specialist code review, and parallel test generation, plus a lightweight /fix workflow and documentation generation from project artifacts.

Standardize AI-generated projects with Ralph autonomous loop, persistent memory, and quality gates. Enforce Git workflow standards, code quality gates via pre-commit/pre-push hooks, and spec-driven task management with OpenSpec format. Manage skills and tasks through CLI commands and an MCP server for Cursor or Claude Code integration.

Review PRs for code quality and consistency, refine ideas into structured goals, build single-file HTML/JS/CSS tools, audit Node.js supply chain security, and manage Val Town projects via CLI.

Automate BMad Method project workflows including code audits, documentation, release management, issue/PR triage, adversarial reviews, LLM prompt auditing, root cause analysis, and skill publishing to Gemini/ChatGPT.

Prepare GNOME Shell extensions for extensions.gnome.org submission: scaffold projects with compliant structure, run automated linting and compliance checks, simulate EGO reviewer audits with rejection scoring, perform code reviews and field tests, and generate readiness reports.

Enforce git workflow conventions, automate GitHub PRs and issues, and harden project security with supply chain audits, code refactoring, and documentation sync. Automate JS/TS project setup with Husky, Biome, commitlint, and run security scans with GitLeaks. Audit Tailwind CSS, file naming, project structure, and dead code for consistent code quality.

Run git test precommit checks and build validation that automatically analyze and fix failures until all commits pass, with iterative testing across every commit in a branch.

Execute Hamster Studio briefs from Claude Code: schedule tasks in parallel waves, run diff-aware QA, perform paranoid code review, and generate engineering retrospectives.

Manage the full Dynamics 365 / Dataverse lifecycle: explore metadata, CRUD records, debug plugins, manage solutions and ALM, audit security, and configure web resources, webhooks, and service bus endpoints — all from Claude.

Migrate Next.js projects to vinext, a Vite-based reimplementation, by scanning compatibility, replacing packages via npm/yarn/pnpm, generating Vite configs, converting to ESM, and configuring deployments for Cloudflare Workers or Nitro.

Configure FeatureDrop toolkit in React apps via Claude Code to add changelogs, onboarding tours, checklists, badges, hotspots, feedback widgets, and surveys optimized for Tailwind or shadcn-ui styling, accelerating product adoption setup.

Author structured spells in a DeFi DSL using Grimoire CLI to fetch real-time market data from Aave V3, Uniswap, Morpho Blue, Pendle, Polymarket, Hyperliquid, and Across, then compile, validate, simulate, and cast them for deterministic agentic execution with explicit AI boundaries.

Automate Brazilian judicial workflows: analyze civil/criminal processes, draft decisions, calculate CPC/CPP deadlines, download PDF case files from PJe courts, extract text via OCR, search CJF jurisprudence, and generate stylized procedural reports in DOCX/PDF.

Scan, triage, and fix software supply chain security risks — assess CI/CD posture, review dependency vulnerabilities, and open remediation PRs using Endor Labs intelligence

Intercepts package install commands across npm, pnpm, pip, go, and cargo to enforce supply-chain risk policies, with a prompt reminder before executing any Bash command to prevent accidental destructive shell operations.

Build and extend FastMCP MCP servers in TypeScript through interactive code generation for custom tools with Zod schemas, static resources, authentication setups including API keys and OAuth, plus patterns for logging, streaming, and merge-friendly import sorting to cut Git conflicts.

Run cheminformatics workflows on SMILES, SMIRKS, and InChI notations via a RDKit-powered WASM CLI: validate structures, repair errors, convert formats, and perform analysis, with structured JSON output for scripting and agent integration. Installs via npm for fast subprocess calls from Node or Bash.

Run structured workflows and CLI commands to diagnose, auto-fix, manually intervene, and verify markdownlint errors (e.g., MD029 ordered lists) in .md files or folders. Reports total scanned files, error types, counts, file names, and line numbers using markdownlint-cli2.

Upload markdown files to Arweave for permanent decentralized storage with automatic provenance tagging including IPFS CID, author name, and app details. Receive ViewBlock transaction URL, direct Arweave link, and IPFS CID for verification and local logging via CLI command or skill invocation.

Securely install, scan, audit, and monitor packages across 14 ecosystems (npm, yarn, pnpm, bun, pip, uv, poetry, cargo, go, gem, composer, nuget, hex, pub) with CVE detection via OSV.dev, auto-fixes, severity triage, dependency audits, and continuous email alerts using INFYNON CLI and agent.

Orchestrate multi-agent teams in isolated Git worktrees for Claude Code to prototype divergent designs, debate trade-offs, perform premortem risk analysis, stress-test vulnerabilities, review code quality/security, and automate full PRD builds from spec decomposition to verified merging.

Orchestrate multi-agent autonomous development workflows: decompose features into testable stories, run parallel SDLC pipelines with isolated git worktrees, fix build/lint/TypeScript errors, execute TDD cycles, and inspect swarm state—all while enforcing tool-use policies and persisting session context via reflective memories.

Lint, test, and scaffold Jython and Playwright projects for Ignition SCADA, including API references for system.* and expression language

Launch a browser-based GUI to assign Zelda BotW and TotK sound effects to Claude Code events like session starts, notifications, tool uses, and subagent actions. Preview sounds, save overrides instantly, and hear them play automatically via Node.js hooks during interactions.

Enforce a structured AI-driven development workflow across idea, research, plan, task, implement, review, and QA stages with Python-based policies, loop-mode discipline, and security hooks for Claude Code sessions.

Scaffold new or migrate existing monday.com code apps with React/Vite frontend and Express/Node backend to serverless/CDN deployment. Run local dev servers with Vite, Express, and Docker MongoDB. Manage production with multi-region deploys, env vars/secrets, cron jobs, alerts, security scans, logs, and status checks via Monday API.

Validate .claude plugin folder structures and rules YAML on writes/edits/bash ops, enforce npm/yarn/pnpm via lockfiles, inject CLAUDE.md project context for discovery, auto-mkdirs, run TSX via npx, and guide Feature-Sliced Design implementation with custom views layer in Next.js apps.

Orchestrate structured AI debates with Toulmin arguments: create debate motions and setups, run multi-turn exchanges between proposition/opposition agents judged by an impartial scorer, and generate reports with analysis, verdicts, and argument graph visualizations.

Generate complete Docus-based documentation sites with search, dark mode, MCP server, and llms.txt integration for Nuxt projects supporting monorepos and various package managers. Review Markdown docs for quality, clarity, SEO, technical correctness, and Docus/Nuxt Content optimizations with prioritized recommendations.

Orchestrate Model Context Protocol (MCP) server lifecycles by installing, removing, listing, and updating tools for GitHub search, SQL queries, and Figma access. Resolve capability gaps through troubleshooting and discover new integrations to extend AI agent capabilities.

Index JS/TS dependency docs and source code locally, then search them with semantic retrieval and reranking — all offline via Docker containers.

Orchestrate a complete feature development lifecycle from codebase analysis and TDD test design through multi-agent code review, UI verification, and PR creation, with state persistence to resume interrupted workflows