No user-submitted related servers found.
Scored across 61 tools
Many tools are clearly isolated by prefix and detailed when/not-when guidance, but there are multiple easy-to-confuse pairs: browser_click vs session_click, fetch_page vs session_text, research vs research_start, citation_pack vs citation_report, and even service_route vs tool_hint. The descriptions reduce ambiguity, but with 61 tools an agent still faces non-trivial selection risk.
Names are consistently lowercase snake_case and mostly follow a verb_noun or prefixed action pattern like session_navigate, research_start, extract_links, and map_site. A few bare nouns like rss, sitemap, snapshot, ping, and stats break the pattern, but they are readable and predictable enough.
61 tools is far too many for a single MCP server, even for a broad browser-research assistant. The 23 session_* tools, duplicated browser_* one-shot equivalents, and meta-tools like service_route, tool_hint, stats, and tool_usage could be consolidated or split into separate servers.
The tool surface covers the full research workflow: search, deep research campaigns, resuming, status, reporting, digesting, citation packs, document reading, extraction, table extraction, browsing, live sessions, screenshots, proxies, profiles, and export. There are no obvious dead ends; lifecycle tools exist for both research campaigns and browser sessions.