Инструменты

Orchestrates multi-agent teams using Claude Code's experimental Agent Teams feature: parallel multi-angle code review, competing-hypothesis debugging and feature work split by file ownership. Enable CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS first.

Plugin for building LLM apps: LangGraph agents with memory and tool use, hybrid-search RAG, vector index tuning, prompt optimization and LLM evaluation. Includes 3 subagents and 8 skills.

MLOps plugin from the wshobson/agents marketplace: data scientist, ML engineer and MLOps engineer subagents and a /ml-pipeline command that runs the pipeline in phases from data analysis and training to deployment and monitoring, plus ML pipeline and recommender system skills.

Official Anthropic skill for cleaning messy Excel exports: trims spaces, standardizes case and dates, converts text numbers, dedupes, and flags mixed types, garbled text and errors. Keeps originals via helper-column formulas and confirms each step. Works in Claude for Excel; local files need Python.

Send files and folders between any two computers with one command: end-to-end encryption (PAKE), direct peer-to-peer with relay fallback, resumable transfers and proxy support, with no server setup or port forwarding.

Cross-platform shell built around structured data: pipelines pass tables instead of plain text, and you can open and query JSON, CSV, TOML and more directly. Default builds include nu --mcp, which serves it to agents as an MCP server.

Converts PDFs, images, PPTX, DOCX, XLSX, EPUB and more into Markdown, JSON or HTML, preserving tables, formulas and layout, with optional LLM help for higher accuracy. Needs Python 3.10+ and PyTorch; large-scale commercial use of the model weights needs a paid license.

Community collection maintained under GitHub's official org with hundreds of skills following the Agent Skills spec, plus custom agents, instructions and plugins across many languages, frameworks and workflows. Content is third-party; install single skills with gh skills install.

PostHog's official hosted MCP: lets agents query product analytics with HogQL, manage feature flags and A/B experiments, and investigate errors and session replays. Sign in to PostHog to authorize; tool calls are free, though a few LLM-powered tools count toward AI usage.

Drafts self-reviews, peer reviews, 360 and upward feedback in STAR structure with concrete examples instead of filler, as Markdown drafts it never submits. With Microsoft WorkIQ it gathers your contributions from Microsoft 365 mail and meetings; otherwise it works from your notes.

Community skill collection in GitHub's official repo: 400+ skills following the Agent Skills spec, covering codebase onboarding, testing, cloud and DevOps, agent security and more. Install individual skills with gh skills install.

Meeting minutes skill from GitHub's awesome-copilot: turns notes, transcripts or speech-to-text output into structured minutes with attendees, agenda, decisions and action items with owners and due dates, asking when details are missing. Best for internal meetings under an hour.

Community collection under GitHub's official org: Copilot custom agents, instructions, skills, hooks and plugins. It's also the default plugin marketplace in Copilot CLI and VS Code; install with copilot plugin install. Content is third-party, so review before installing.

Multi-agent orchestration plugin for Claude Code: /team coordinates subagents through a staged plan, requirements, execution, verification and fix pipeline, /autopilot drives tasks autonomously, and it can dispatch Codex, Gemini and other CLIs in tmux to work in parallel.

A smarter cd that remembers your directories, ranking them by frequency and recency so you can jump with a fragment of the name. Add zoxide init to your shell config after installing.

Static analysis for shell scripts: flags syntax errors and common bash/sh pitfalls such as quoting and variable expansion, with suggested fixes. Good for checking agent-generated scripts and easy to add to CI.

ByteDance's open-source multimodal agent CLI: operates browsers via vision or the DOM and calls terminal, file and MCP tools to complete tasks, with a Web UI and headless server mode. Needs Node.js 22+ and a model API key.

Browser automation MCP server open-sourced by ByteDance in the UI-TARS-desktop repo: drives pages through Puppeteer accessibility data with an optional vision mode, running locally over stdio or remotely over SSE or Streamable HTTP.

All-in-one security scanner for container images, filesystems, Git repos and Kubernetes clusters: finds known vulnerabilities, IaC misconfigurations, leaked secrets and software licenses, and generates SBOMs. Open source, maintained by Aqua Security.

Define multi-container apps in compose.yaml and bring up services, databases, caches and the rest with one docker compose up, handy for agents setting up local dev and test dependencies. Maintained by Docker; included with Docker Desktop on macOS.

Detects your CPU, memory and GPU and scores open models on whether they run, speed, quality and context, then works with local runtimes like Ollama, llama.cpp and MLX to download and benchmark them. recommend --json gives agent-friendly output.

Local model gateway that routes requests from Claude Code, Codex and other coding agents to providers like DeepSeek, Kimi, Gemini and OpenRouter, with failover, rule-based routing and request logs. Bring your own API keys.

Alibaba's open-source Java production diagnostics tool: inspect method arguments and timing, threads and GC, class loading and decompiled code, and generate flame graphs without changing code or restarting the JVM. Supports JDK 8+ via an interactive CLI.

Headless browser written from scratch in Zig for AI agents and automation, using less memory and running faster than headless Chrome in official benchmarks. fetch turns pages into Markdown, and it can serve CDP or MCP. Telemetry is on by default (an env var disables it); Windows needs WSL.