AI-assisted administration#
Ask in chat. YAADMIN selects from enabled registered tools using their names, descriptions and schemas, then follows runtime execution checks. Actual capabilities depend on the tools you connect.
Modules and external tools#
Seven built-in modules cover Action1, read-only Active Directory, Codex, files, MCP, read-only 1C and target resolution. External MCP discovery adds server-specific tools. Explore the current modules.
Access at the capability level#
Roles, permission effects and scopes govern execution. Each discovered MCP tool receives its own permission. Users can have different execution access; denied tools are not proven to be hidden from AI catalogs.
Approval-aware operations#
Review fingerprint-bound requests before approved execution. Routine actions may run automatically under policy/RBAC. Approvals are in memory and allow self-approval by default. Understand the boundaries.
Live Activity and auditability#
Follow tool and approval events, inspect results and use administrator audit visibility. Stop requests cancel processing without promising to reverse external actions already submitted.
An everyday workspace#
Owner-scoped history/folders/search, saved prompt commands, files, tables and Excel export support repeatable administration. English, Ukrainian and Russian preferences are per user.
Local and AD identity#
Local/AD authentication and user lifecycle exist, with a protected local recovery administrator. Credential-flow evidence is partly redacted; AD configuration remains deployment-specific.
Extend deliberately#
External packages can register tools without generic Core edits. Read the SDK's actual fields and parser limitations before building. Create a module.