Related Servers

The core actions (navigate, click, type_text, snapshot) are clearly distinct, and descriptions actively steer choices (prefer snapshot over screenshot). However, the set is riddled with near-duplicate convenience wrappers — navigate vs navigate_and_snapshot, scroll vs camofox_scroll_element vs scroll_and_snapshot vs camofox_scroll_element_and_snapshot, click vs batch_click, type_text vs type_and_submit vs fill_form — and extraction tools (snapshot, get_page_html, query_selector, evaluate_js, extract_structured) require reading descriptions to separate. load_profile vs import_cookies also overlap in purpose.

Everything is snake_case, which keeps it readable, and most names follow verb_noun (create_tab, list_downloads, delete_profile, batch_click). But an arbitrary subset is prefixed with camofox_ (camofox_hover, camofox_wait_for, camofox_evaluate_js, camofox_close_session) while their direct siblings (click, wait_for_text, snapshot) are not, and several are bare verbs/nouns (navigate, snapshot, refresh), so the pattern breaks within the same functional category.

47 tools is excessive for a browser automation server, and much of the bloat comes from combined convenience variants that duplicate an existing primitive plus snapshot. Many tools earn their place (tabs, downloads, profiles, cookies, extraction), but the surface is heavy enough that an agent must triage before acting.

Coverage is broad: tab lifecycle, navigation/history, interaction, multiple extraction paths, downloads, blob resolution, profile and cookie persistence, geo presets, search macros, display mode, and stats. Minor gaps remain — no file upload, no explicit dropdown/select_option, and no cookie listing/clearing beyond import — but agents can work around these via JS evaluation.