Skip to main content
Gologin AI Skills are plug-and-play extensions for Claude Code that give AI agents web scraping and browser automation capabilities. Each skill wraps a Gologin CLI tool and exposes it as a set of agent-callable functions.

Available skills

Web Access Skill

Default skill for known-site reading, extraction, crawling, monitoring, and hybrid scrape-then-browse tasks.

Agent Browser Skill

Live browser sessions in Gologin Cloud Browser or local Orbita: ref-based interaction, screenshots, PDFs, profile warmup, and runbooks.

Scraping Skill

Scraping-only skill wrapping the Web Unlocker SDK. Smallest footprint.
The Local Agent Browser skill is deprecated. The Agent Browser skill now handles local Orbita profiles through gologin-agent-browser --runtime local.

Install

All Gologin skills install from the GologinLabs/agent-skills repository:
Each skill requires its underlying CLI to be installed:

Which skill to use


Web Access Skill

Install: npx skills add GologinLabs/agent-skills@gologin-web-access-skill The default and most versatile skill. Delegates to gologin-web-access CLI. Scraping tools: scrape_url, read_page, scrape_markdown, scrape_text, scrape_json, batch_scrape, batch_extract, search_web, map_site, crawl_site, crawl_site_async, extract_structured, track_changes, batch_track_changes, parse_document Browser tools: browser_open, browser_search, browser_snapshot, browser_click, browser_type, browser_screenshot, browser_close, browser_sessions, browser_current Built-in workflows: browser-ref-loop, hybrid-read-then-interact, search-discover-and-scrape, lead-enrichment, competitive-monitoring, docs-ingestion, geo-testing, login-and-navigate-dashboard See Web Access CLI reference for full command details.

Agent Browser Skill

Install: npx skills add GologinLabs/agent-skills@gologin-agent-browser-skill Browser-only skill for live sessions in Gologin Cloud Browser (default) or in local Orbita profiles with --runtime local. Use when you need interactive browsing without scraping. Capabilities: doctor, open, sessions, current, snapshot, click, type, fill, find, get, tabs, tabopen, tabfocus, tabclose, cookies, cookies-import, storage-export, storage-import, eval, back, forward, reload, upload, screenshot, pdf Local runtime adds: profile management (profiles, profile-create, profile-import, profile-update, profile-sync, profile-delete), runbooks (run, batch), jobs (jobs, job), use-case templates (linkedin, ads, smm, scraping, geo), and diagnostics (doctor --use-case --check-proxy). See Agent Browser CLI reference and Local runtime for full command details.

Scraping Skill

Install: npx skills add GologinLabs/agent-skills@gologin-scraping-skill The smallest skill — scraping only through the Web Unlocker SDK. No browser session, no daemon. Capabilities: Raw HTML (scrape), text (text), markdown (markdown), structured metadata (json). SDK methods: scrapeRaw(), scrape(), scrapeText(), scrapeMarkdown(), scrapeJSON(), batchScrape(). Use when the task is scraping-only and you want a smaller dependency footprint. See Web Unlocker SDK reference for API details.

Local Agent Browser Skill (deprecated)

If your agent still has gologin-local-agent-browser-skill, it only points the agent to the Agent Browser skill. Install gologin-agent-browser-skill instead: it covers local Orbita profiles, warmup, login flows, runbooks, and batch jobs.

What you can ask your agent to do

Once a skill is installed, you talk to Claude Code in natural language. Here’s what each skill enables:

With Web Access Skill

  • “Read the pricing page at competitor.com and summarize the plans”
  • “Scrape these 15 URLs and extract the product name and price from each”
  • “Crawl the docs at docs.example.com and give me a summary of each section”
  • “Search for ‘best CRM tools 2026’ and scrape the top 5 results”
  • “Check if competitor.com/pricing changed since yesterday”
  • “Extract all blog article titles and URLs from this site”
  • “Open example.com in a browser, log in with these credentials, and screenshot the dashboard”

With Agent Browser Skill

  • “Open the admin dashboard, go to Settings, and screenshot the current config”
  • “Log into the staging environment and check if the new feature is deployed”
  • “Fill out the signup form on example.com with these test details”
  • “Navigate to the reports page, download the monthly export, and screenshot the summary”
  • “Check what iphey.com shows for this browser profile”

With Scraping Skill

  • “Scrape this URL and give me the text content”
  • “Get the title, description, and headings from these 10 pages”
  • “Fetch this page as markdown so I can analyze it”

With Agent Browser Skill (local runtime)

  • “Create a new LinkedIn profile with a US proxy and warm it up”
  • “Open my Reddit account, scroll through the feed, and close it”
  • “Run the warmup runbook across all 20 profiles with concurrency 5”
  • “Check the proxy health on my Facebook ad profiles”
  • “Log into this marketplace account and screenshot the seller dashboard”

Example workflows

Competitive research (Web Access Skill)

  1. Agent searches for competitors: search_web("saas project management tools")
  2. Agent maps a competitor site: map_site("https://competitor.com")
  3. Agent batch-scrapes key pages: batch_scrape(pricing, features, about)
  4. Agent feeds content to Claude for analysis and comparison

Account warmup (Agent Browser Skill, local runtime)

  1. Agent creates profiles: profile-create "Account 1" --template linkedin
  2. Agent runs warmup runbook: batch ./warmup.json --targets ./profiles.json
  3. Agent checks job results: jobs --kind batch --status ok
  4. Agent reports which profiles are ready

Dashboard monitoring (Agent Browser Skill)

  1. Agent opens dashboard: open https://dashboard.example.com --profile saved_login
  2. Agent reads the page: snapshot
  3. Agent navigates to metrics: click @e5 (Reports link)
  4. Agent screenshots the report: screenshot ./daily-metrics.png
  5. Agent closes: close
  6. Agent summarizes the metrics from the snapshot data

Mandatory preflight

Before running any task, skills classify the work:
  1. One known site or broad multi-source research? — determines if scraping or search is needed
  2. Read-only extraction, monitoring, or interactive browser work? — determines scraping vs browser
  3. Only Web Unlocker, or both Web Unlocker and Cloud Browser? — determines credential needs
  4. Geo-sensitive or blocked enough for Gologin end to end? — determines proxy/profile requirements