Related Servers

Multiple ecosystem_* tools overlap heavily: ecosystem_scan vs ecosystem_scan_periodic vs ecosystem_index_update vs ecosystem_refresh vs ecosystem_quick_setup all touch scanning/indexing; ecosystem_search vs ecosystem_search_by_capability; ecosystem_summary_by_tag vs ecosystem_summary_top_n vs ecosystem_summary_weekly; and ecosystem_deep_review_request vs ecosystem_deep_review_request_batch vs ecosystem_deep_review_status vs ecosystem_deep_review_cancel. Task, meeting, and memory groups are mostly distinct, but the ecosystem cluster and broad search tools like unified_search vs memory_search create real misselection risk.

All tool names use snake_case, and many carry domain prefixes (project_, team_, task_, meeting_, memory_, channel_, ecosystem_). However action order varies: noun_verb (project_create, team_list, meeting_create), verb_noun (unified_search, find_skill, verify_completion), and noun-only phrases (prompt_effectiveness, usage_attribution, context_resolve). It is readable but not a single predictable pattern.

116 tools is extreme for an MCP server, well beyond the 50+ threshold for a poor score. Although the AI Team OS domain is broad, this many tools will overwhelm context, increase selection latency, and make maintenance difficult. Many ecosystem_* and memory_* tools could be consolidated or gated behind a smaller surface.

The surface covers the orchestration domain extensively: projects, teams, agents, tasks, memos, meetings, memory, channels, ecosystem pipeline, workflows, config, and health checks. Minor gaps exist, such as no standalone task_delete (only via project_delete), no report update/delete, and no event deletion, but these are workaroundable. Overall it is highly complete, with only minor lifecycle omissions.