Scored across 30 tools
Most tools are cleanly separated by resource prefix (memory_, notes_, tasks_, project_, session_, gardener_) and the descriptions clearly state when to prefer each one. The main ambiguity risk is the write/edit/append cluster within memories and notes, where the boundaries are subtle despite thorough descriptions.
The dominant <resource>_<action> pattern (memory_write, notes_read, tasks_claim, project_list, session_start) makes the set predictable. A few outliers break it: capture_url and recall use verb-first forms, usage_summary is a noun phrase, and favorite_set inverts the usual action-object ordering.
30 tools is well past the 25+ threshold and the set feels inflated by parallel exact-search/replace, whole-body-update, and append variants for both memories and notes, plus multiple gardener planning entry points. The broad domain explains some of the count, but an agent has to navigate a lot of near-synonyms.
Memory, note, and task lifecycles are essentially complete, and recall/capture/favorite cover retrieval and intake. Projects are only create/list with no update, rename, or delete path, and sessions have start/update/end but no way to inspect a past session beyond aggregate usage_summary.