Related Servers

The six search tools (search_entries, search_content, search_natural, search_by_name, plus the name/path resolvers) have overlapping purposes despite descriptions that mostly route between them; the repeated warnings about which to use ('prefer search_content when...', 'search_entries = raw query') signal genuine ambiguity that an agent can misfire on. Task tools are also muddled: get_task_status, wait_for_task, and task_update all do the same underlying job with different timeout semantics.

Most tools follow verb_noun (entry_get, field_values_get, folder_list, document_get_edoc) but verbs are inconsistently placed: entry_get vs get_field_values vs document_find_duplicates vs task_wait. The description of task_update explicitly acknowledges the naming breaks its own convention (a '_update' that is not a write). Readable, but not a predictable pattern.

23 tools is heavy for what is essentially a document-repository surface, and the count is inflated by redundant aggregation: four search tools, three task-status tools (get_task_status, wait_for_task, task_update covering the same payload), and near-duplicate definition listers. Several tools collapse into one another, so the set carries more names than distinct operations.

Coverage is broad: entry read/write, folder listing, document text/edoc/image retrieval, duplicate detection, field/tag/template/link/audit metadata, and async task tracking. Missing write-side operations (create/move/rename/delete are implied by siblings like delete_entry but not all listed here) and no update for entry attributes, but the core CRUD and lifecycle are largely present.