Инструменты

CLI and library for evaluating and red-teaming LLM apps: compare prompts and model outputs in bulk, scan for security vulnerabilities and plug into CI/CD, with evals running locally. Now part of OpenAI and still open source under MIT.

Removes image backgrounds with one command: single images, whole folders or watched directories, mask-only output and switchable models like U2Net and SAM. Can also run as an HTTP server or process video frames with FFmpeg. Default models run locally.

Official plugin for Pascal, an open-source 3D architecture editor: lets agents create, edit and validate 3D models of rooms and floors over MCP and check furniture placement for collisions or blocked doors. Local use needs the Pascal CLI (Node.js 22.13+); hosted projects need an API key.

SVG optimizer CLI and Node.js library: shrinks SVGs by removing editor metadata, comments, hidden elements and redundant attributes, processes whole directories recursively, and gives fine control through plugin config.

ML-based video upscaling and frame interpolation using Real-ESRGAN, Real-CUGAN, RIFE and other models to upscale old low-res video or raise frame rates; video2x handles one video per run and is scriptable. Needs a Vulkan GPU; Windows, Linux and Docker builds, no native macOS.

Google's official skill library and plugin marketplace: nearly 150 skills focused on Google Cloud (GKE, BigQuery, Cloud Run, IAM and more), plus 17 product plugins pairing skills with MCP for Cloud SQL, Spanner, AlloyDB, BigQuery, GCS and others.

Manages BigQuery datasets, tables and jobs with the bq and gcloud CLIs, runs SQL queries and loads data, with references for client libraries, MCP and Terraform. Google's official skill; requires a Google Cloud project.

Google's official skill library, focused on Google Cloud: 100+ skills for gcloud, Cloud Run, GKE, BigQuery, IAM, monitoring and logging, and the Well-Architected Framework, plus Ads and Analytics.

Industrial-grade open-source speech recognition toolkit from Alibaba's ModelScope, with an agent-friendly funasr command that turns audio into text, JSON or SRT subtitles in one step, with speaker diarization and timestamps. Strong on Chinese. Requires PyTorch.

Command-line tool for bulk-downloading galleries and posts from many image hosts and social platforms, with flexible file naming and config, and yt-dlp integration for video. Main development has moved to Codeberg; the GitHub repo stays in sync.

AI2's tool for converting PDFs and scans to Markdown with a vision LLM, handling equations, tables, handwriting and multi-column layouts, suited to bulk digitizing papers and archives. Local runs need an NVIDIA GPU with 12GB+ VRAM; otherwise use a paid OpenAI-compatible remote endpoint.

Browser control from the browser-use team: an editable CDP connection lets agents drive the Chrome you are already using, writing missing helper functions as they go and keeping them for reuse. Also offers an MCP server. Allow remote debugging in Chrome on first use.

Lets Claude Code push ML research forward unattended: literature review, topic selection, experiment deployment, cross-model review loops and paper writing through rebuttal, in 80+ skills. Codex acts as the independent reviewer by default; other model pairings work too.

Context and harness engineering skills covering context compression and optimization, multi-agent architectures, memory systems, tool design and LLM-as-judge evaluation, activated automatically by task. Aimed at building production-grade agent systems.

Context and harness engineering skills: context fundamentals, degradation diagnosis and compression, multi-agent architectures, memory systems, tool design, LLM-as-judge evaluation and self-improvement loops. Focused on principles and design patterns, not tied to any agent platform.

Official Lark (Feishu) skill to view, export and edit whiteboards in Lark docs: write Mermaid, PlantUML or SVG into a board for flowcharts or org charts, or export boards as images or SVG. New boards are created with lark-doc. Needs lark-cli, Node.js and auth; install with lark-shared.

Veteran image processing CLI suite that converts, resizes, crops, composites and batch-processes JPEG, PNG, GIF, TIFF, PDF and many more formats, handy for agents handling images in scripts.

GSAP's official animation skills: teach agents to use the core API, timelines, ScrollTrigger and plugins correctly, including React, Vue and Svelte integration and animation performance. Maintained by GreenSock; all plugins are free.

Alibaba's open-source offline data sync tool: JSON job configs batch-sync between heterogeneous sources like MySQL, Oracle, PostgreSQL, Hive, HDFS and MaxCompute, run with python datax.py. Needs JDK 1.8+ and Python; download and unpack the official toolkit.

Helps support teams categorize and prioritize tickets, emails or chat logs and suggest next steps; drafts replies, organizes repro steps and reminds you to redact customer data before sharing. Works with exports from Zendesk, Intercom and others or pasted chats; no account or API key needed.

Simplifies Figma designs into layout and style data for coding agents, so UIs built from the design are more accurate than working from screenshots. Requires a Figma personal access token.

ModelScope's official training framework for LLMs and multimodal models: the swift command covers pretraining, fine-tuning, RL, inference, evaluation, quantization and deployment for 600+ text and 400+ multimodal models, with Megatron parallelism and a web UI. Usually needs a GPU.

Version datasets and models the way Git versions code: data lives in remote storage like S3 or GCS while versions are tracked in your Git repo. Also offers reproducible pipelines and local experiment tracking, rerunning only the steps affected by a change.

Bundles a whole web page with its CSS, images, fonts and scripts into a single HTML file that opens offline exactly as it looked, good for archiving pages or handing them to an agent for offline analysis. It does not run JavaScript, so pre-render dynamic pages with a headless browser.