
PageMap
A web automation layer built specifically for AI agents that need to navigate and interact with pages without blowing through context windows. Exposes 13 MCP tools including get_page_map, execute_action, fill_form, and tab management. Instead of dumping 50KB accessibility trees like Playwright or read-only markdown conversions, it compresses HTML into 2-5K token structured maps with clickable element references. Auto-detects 16 page types, handles cookie consents and bot barriers, and includes deep extraction for 30+ e-commerce sites. Works as stdio MCP server, Python SDK, or Docker HTTP transport. Reach for this when you're building agents that browse multiple pages in one session or need form interaction beyond simple scraping.
- GitHub stars
- 32
- tools
- 13
- transport
- STDIO
- last updated
- Feb 26, 2026
Tools
13 toolsPublic tool metadata for what this MCP can expose to an agent.
get_page_mapNavigate to a URL and get a full structured map with numbered refs.
Navigate to a URL and get a full structured map with numbered refs.
No parameter schema in public metadata yet.
execute_actionClick, type, select, or hover using a ref number from the last `get_page_map`.
Click, type, select, or hover using a ref number from the last `get_page_map`.
No parameter schema in public metadata yet.
fill_formFill multiple form fields in one call. More efficient than sequential `execute_action` calls.
Fill multiple form fields in one call. More efficient than sequential `execute_action` calls.
No parameter schema in public metadata yet.
get_page_stateCheck current URL and title without a full rebuild. Use after actions that may navigate.
Check current URL and title without a full rebuild. Use after actions that may navigate.
No parameter schema in public metadata yet.
scroll_pageScroll to reveal lazy-loaded content before calling `get_page_map` again.
Scroll to reveal lazy-loaded content before calling `get_page_map` again.
No parameter schema in public metadata yet.
wait_forWait for dynamic content to appear (e.g. after a search or form submit).
Wait for dynamic content to appear (e.g. after a search or form submit).
No parameter schema in public metadata yet.
take_screenshotCapture the visual state when the PageMap alone is ambiguous.
Capture the visual state when the PageMap alone is ambiguous.
No parameter schema in public metadata yet.
navigate_backGo back one step in browser history.
Go back one step in browser history.
No parameter schema in public metadata yet.
open_tabOpen a new browser tab and navigate to a URL.
Open a new browser tab and navigate to a URL.
No parameter schema in public metadata yet.
switch_tabSwitch to a different open tab by index.
Switch to a different open tab by index.
No parameter schema in public metadata yet.
list_tabsList all open tabs with their URLs and titles.
List all open tabs with their URLs and titles.
No parameter schema in public metadata yet.
close_tabClose a tab by index.
Close a tab by index.
No parameter schema in public metadata yet.
batch_get_page_mapFetch multiple URLs in parallel. Use for comparison tasks.
Fetch multiple URLs in parallel. Use for comparison tasks.
No parameter schema in public metadata yet.
Sponsors
Readme
PageMap
PageMap converts raw HTML (100K+ tokens) into structured, AI-readable page maps (2-5K tokens) — a 97% token reduction. It works as an MCP server, Python SDK, and CLI, supporting 16 page types and 30+ e-commerce sites. Agents can read, click, type, and navigate any web page.
"Give your agent eyes and hands on the web."
Details
- Repository
- github.com/retio-ai/retio-pagemap
- Categories
- Web & Browser Automation, Search & Web Crawling
- Package
retio-pagemap- Transport
- STDIO
- Registry
- MCP registry, active