No user-submitted related servers found.
Scored across 21 tools
Most tools have clearly distinct roles, with a sensible layering of observe (state, snapshot, query, screenshot), act (click, hover, type, key, scroll, navigate), and extend (eval, script, css). A few pairs could be confused at a glance — snapshot vs query (list vs detail), eval vs script (one-shot vs persistent injection), state vs map (live vs crawled) — but the descriptions draw the boundaries explicitly.
All names are single lowercase snake_case tokens (wait_for, lcu_endpoints), with no camelCase mixing. The set mixes nouns for read-only inspectors (console, network, metrics, map) with verbs for actions (click, hover, type, navigate), which is a defensible and readable pattern rather than a chaotic one.
21 tools is on the heavy side, but each covers a genuinely separate capability of client automation (input, DOM inspection, navigation, LCU access, JS injection, diagnostics). Nothing looks like a redundant duplicate of another tool, so the count is justified by scope even if it sits near the upper bound.
The surface covers the full loop: inspect state, take snapshots, act with real input, wait, verify via console/network/metrics, and extend via eval/script/css plus direct LCU calls and an endpoint catalog. Minor gaps exist — no drag/double-click or multi-step form fill, no console/network clearing or filtering — but these are workaroundable via eval or repeated calls.