You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
A Model Context Protocol (MCP) server that exposes GLPI (IT Service
Management) to AI assistants like Claude.
v3 is a foundations-and-coverage overhaul on top of v2. See
CHANGELOG.md for the full list of changes; the rest of this
README documents what's exposed today.
What's in v3
Solid foundations: unified HTTP layer with auto re-authentication on 401,
structured errors, retry on 5xx.
Real search: multi-criteria with AND / OR / AND NOT / OR NOT,
forcedisplay, pagination, fetch_all, dedicated count probe.
Dynamic field mapping: /listSearchOptions/{itemtype} is cached so
field_id ↔ name translations stay valid across GLPI versions.
High-level reporting: glpi_search_tickets accepts friendly params,
glpi_tickets_stats_by ventilates counts by status / category / technician /
entity / month.
glpi_list_*, glpi_get_*, and glpi_create_* where the GLPI API allows it.
glpi_search_knowbase performs a free-text search on the title (field id
resolved dynamically — no longer hard-coded).
Users, groups, categories, entities
glpi_list_users filters active users via the search endpoint (not
searchText); glpi_create_user/glpi_create_group/glpi_add_user_to_group
cover provisioning.
Statistics
Tool
Description
glpi_get_ticket_stats
Counts by status (optional entity / date filters)
glpi_get_asset_stats
Total counts per asset type
glpi_tickets_stats_by
Ventilation by status / category / technician / entity / month, optional period