skills.rest
nbiish
nbiishCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

skyvern

Automate browser workflows, form filling, and data extraction using self-hosted Skyvern with Playwright.

Updated Sep 2, 2026
Search Tags:#data extraction#browser automation#web scraping#playwright#self-hosted#litellm#skyvern
One-click install
npx skills add https://github.com/nbiish/tts-mcp --skill skyvern-nbiish
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: skyvern
Source: https://github.com/nbiish/tts-mcp/tree/main/.agents/skills/skyvern
Command: npx skills add https://github.com/nbiish/tts-mcp --skill skyvern-nbiish

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires skyvern.

What problem does it solve? Web automation built on brittle CSS/XPath selectors breaks whenever a site changes its layout, and hosted automation APIs force you to send data through third-party clouds. This Skill guides you through deploying and operating a fully self-hosted Skyvern instance that uses vision-based LLM agents to navigate websites semantically, with all LLM inference routed through your own providers. ## Core Features & Use Cases - Self-Hosted Deployment: Install via pip, Docker Compose, or Kubernetes, and configure LLM providers through LiteLLM (OpenRouter, Ollama, OpenAI-compatible endpoints like ZenMux, Z.AI, or Nebius). - Vision-Based Browser Automation: Run tasks and multi-step workflows using the Planner→Agent→Validator engine with Playwright, without writing CSS or XPath selectors. - SDK, REST API, and MCP Integration: Drive automation from the Python or TypeScript SDK, call the local REST API, or expose Skyvern as an MCP server for Claude Code, Cursor, and Windsurf. - Use Case: Build a workflow that logs into a vendor portal with stored credentials, downloads all invoices newer than January 1st, extracts structured line items via a JSON schema, and emails the results. ## Quick Start Set up a self-hosted Skyvern instance with OpenRouter as the LLM provider and run a task that extracts the top post from Hacker News.

Frequently Asked Questions about skyvern

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I self-host Skyvern for browser automation?▼

Install with pip install skyvern and run skyvern quickstart, which configures the database, LLM provider, browser mode, and local API key. Alternatively, clone the GitHub repository, edit .env, and run docker compose up -d to get the UI at localhost:8080.

How do I configure Skyvern with OpenRouter or Ollama?▼

Set ENABLE_OPENROUTER=true with OPENROUTER_API_KEY and OPENROUTER_MODEL, or ENABLE_OLLAMA=true with OLLAMA_SERVER_URL and OLLAMA_MODEL in your .env file. All LLM calls route through LiteLLM, and LLM_KEY selects the registered model config.

Can I use Skyvern with any OpenAI-compatible API endpoint?▼

Yes, set ENABLE_OPENAI_COMPATIBLE=true with OPENAI_COMPATIBLE_API_KEY, OPENAI_COMPATIBLE_API_BASE, and OPENAI_COMPATIBLE_MODEL_NAME. This works with proxies like ZenMux, Z.AI, Nebius, LM Studio, or vLLM, and you can enable vision via OPENAI_COMPATIBLE_SUPPORTS_VISION.

Does Skyvern require CSS or XPath selectors for automation?▼

No, Skyvern uses computer vision and LLM reasoning to discover page elements semantically, making it resistant to layout changes. The Python SDK still supports hybrid mode where a selector is tried first with an AI prompt as fallback.

Why does Skyvern report LLM Provider NOT provided errors?▼

This usually means the Docker image is outdated or the model is not registered. Pull the latest image, confirm the ENABLE_* flag for your provider is set, and check config_registry.py for valid LLM_KEY values.

What Python version does Skyvern support?▼

Skyvern requires Python 3.11.x or 3.12.x and does not support Python 3.13. If pip install fails with a resolution error, use uv pip install skyvern or upgrade to v1.0.32 or later.

Related Skills

Explore More Software Engineering Skills→
BerriAIBerriAI
77

add-model

Register a new LLM deployment in LiteLLM via POST /model/new and validate routing.

Official
Intermediate
modus-bimodus-bi

data-analyst-expert

Analyze BI datasets to generate Level-4 insights with quantified recommendations.

Official
Advanced
Elysium-ArcElysium-Arc
2

webhook_retry

Schedule webhook retries with exponential backoff and circuit breakers.

Official
Intermediate
SilentFlowerSilentFlower

craft-rpa

Record real browser interactions and network traffic, then convert sessions into RPA-ready markdown traces.

Community
Advanced