No user-submitted related servers found.
Scored across 69 tools
The descriptions are exhaustively cross-referenced, which helps, but the 69-tool set still has genuinely crowded areas: four screenshot variants, four read paths (read_page, extract, extract_table, query_dom, find_text), and monitor_network vs network_rules vs track_events vs read_console all need careful reading to separate. An agent can usually pick right thanks to the 'use X instead' notes, but the boundaries are not self-evident from names alone.
Most tools follow snake_case verb_noun (get_tabs, create_tab, set_storage, full_page_screenshot), but there is a large minority of bare verbs and noun-only names (navigate, extract, click, hover, scroll, assert, watch, audit, handoff, clipboard, animations, frames, lighthouse). The convention is readable but only loosely followed.
69 tools is far beyond a comfortable set for an agent, and the existence of an opt-in 'caps' mechanism (audits, visual, perf, network, storage, dom, files) confirms the surface was split because it is too large. Splitting into cap groups is the right mitigation, but as a flat tool list it is heavy.
Coverage is exceptionally thorough for a browser bridge: navigation, tabs/windows, DOM read/write, forms, dialogs, storage, network interception, performance/audit, screenshots, downloads, WebMCP, and human handoff. Almost no plausible browser-automation operation is missing.