Video & Animation

browser-use is an open-source Python framework for letting AI agents interact with websites through a real browser. It supports local execution, optional cloud browsers, custom tools, and a CLI for persistent browser sessions.

Use browser-use to turn natural-language web tasks into repeatable browser automation backed by Playwright and agent loops. This skill helps an agent open sites, inspect page state, click, type, extract data, and recover from common UI changes with a real automation framework instead of brittle one-off scripts.

Automates browser tasks with Browser Use, the open-source library that connects LLM reasoning to Playwright-driven web actions. Useful for navigating sites, filling forms, extracting structured page data, and running agentic browser workflows with screenshots and stateful sessions.

Browser Use is an open source browser automation framework for AI agents that turns websites into controllable interfaces for multi-step tasks. It combines a Python SDK, browser orchestration, and model integrations so agents can navigate, extract data, and complete workflows in real browsers.

Records and replays browser sessions using rrweb recording library with DOM mutation serialization. Integrates with LogRocket API and FullStory Data Export API for session analytics.

Index local coding-agent sessions into a searchable SQLite-backed view so you can inspect usage, compare failures, and recover prior context quickly.

Expose browser-use web automation through an MCP server so Cursor or another MCP client can operate websites with browser state returned to the agent.

broot is a Rust-based terminal tool that provides a new way to see and navigate directory trees. It offers fuzzy search, file preview, git status integration, and an interactive tree view that adapts to your terminal size.

Store credentials once, then inject them into outbound agent requests at runtime so agents can call services without receiving raw secrets.

Uses htmltest to crawl generated documentation or static site output, detect broken internal and external links, and return a link-focused validation report before deploy. This is a narrow docs QA skill for agents validating already-built HTML, not a generic site generator or crawler listing.

Bright Data Web MCP gives MCP-compatible agents live web search, scraping, and optional browser-automation access through Bright Data’s web-access platform. It is built for agents that need current information, anti-bot resilience, and structured extraction without maintaining their own scraping stack.

Let operators control local Claude Code, Codex, Cursor, Gemini CLI, and other coding agents from Slack, Discord, Telegram, Feishu/Lark, DingTalk, LINE, and WeChat Work.

Runs an existing coding agent behind Slack, Telegram, Discord, Feishu, DingTalk, WeCom, WeChat, or HTTP so teams can invoke real agent sessions from chat without rebuilding the agent stack.

Use ZCF to initialize, update, and standardize Claude Code or Codex environments with prompts, workflows, MCP services, API settings, and backups.

Start, watch, and restart a local stack of app processes together instead of juggling many terminals and ad hoc shell loops.

An agent skill that uses BlurHash, the compact image placeholder representation algorithm by Wolt, to generate and decode tiny hash strings that represent blurred previews of images. Enables instant placeholder rendering in web and mobile applications while full images load.

Add a runtime guard that evaluates agent actions, blocks dangerous commands or secret exposure, and audits new skills before they run.

Scan staged changes, commits, or repositories for secrets before they leave the workstation or CI job, instead of relying on a later platform-side catch.

Uses mitmproxy2swagger to turn captured mitmproxy flows or HAR exports into a first-pass OpenAPI document that an agent can refine, validate, and hand off to downstream tooling. Invoke it when the API already exists but the contract does not, and you need a concrete spec before generating clients, mocks, tests, or reviewable docs.

Use CC Safety Net when coding-agent CLIs need pre-execution hooks that block destructive commands, secret access, and unsafe file operations before tools run.

Add hard pre-execution guardrails to Claude Code so destructive shell commands are blocked before an agent can run them.

Put an inline firewall and containment layer in front of agent network traffic, tool calls, and MCP traffic before you trust an agent with local secrets.

Generates Blender Python (bpy) scripts that programmatically create Geometry Nodes modifier trees, using the node_groups API and GeometryNodeTree interface for parametric 3D asset generation.

Automates migration from ESLint and Prettier to Biome (formerly Rome) by parsing .eslintrc and .prettierrc configs, mapping rules to biome.json equivalents, and running biome check --apply for bulk reformatting.