No user-submitted related servers found.
Scored across 29 tools
Tools are grouped by clear namespaces and each has a distinct job: navigation (open/goto/back), interaction (click/type/wait/dismiss), inspection (read/screenshot/tabs), memory (recall/record/rate/stats), atlas lookup (find/site/guard) and sharing (preview/upload/sync). The closest pairs — browser_open vs browser_goto (new tab vs same episode), browser_close vs browser_close_mine, memory_record vs memory_rate, lesson_record vs map_learned — are each differentiated in their descriptions, though memory/lesson/atlas tools still require careful reading to keep apart.
The dominant pattern is namespace-prefixed snake_case (browser_*, memory_*, atlas_*, share_*) with verb-last or verb-first sub-names, which is predictable and readable. Only a few strays break the pattern: lesson_record and map_learned are noun_verb while their peers are verb_noun, and login_sites is noun_noun with no namespace, so the shape is mostly but not fully uniform.
29 tools is on the heavy side for one server, and a few are mergeable: atlas_site is largely a detail view of atlas_find results, browser_close and browser_close_mine could be one tool with a scope argument, and memory_stats is a thin reporting wrapper. The breadth (browser control plus memory plus atlas plus sharing) does justify many tools, so it is heavy rather than bloated, but it sits just past the comfortable range.
The browser lifecycle is well covered end to end: launch, tabs, open/goto/back, read with outlines and offsets, click, type, wait, dismiss, batched run, screenshot, close with verdict, plus login and remembered credentials. Remaining gaps are minor but real — no scroll or raw key-press tool, no file upload or download handling, and no explicit select/option control — all of which an agent would have to improvise around.