NeoCRM AI Agent Operations Kit

What problem it solves

When an AI agent operates inside a CRM, the hard part is not sending messages. It is knowing how to query, mutate, and link entities such as accounts, leads, opportunities, schedules, and public pools without making unsafe assumptions. Ad-hoc command stitching can cause object mismatches, permission mistakes, and missing pre-visit checks. The NeoCRM Skill Pack separates agent behavior, shared operations, and scenario-specific extensions into scannable skill directories, so the model reads capability boundaries instead of raw command lists.

How the skill works

The platform scans subdirectories and loads each skill. nc-base defines general AI agent behavior and is a dependency for other skills. nc-common covers cross-entity standard operations such as query, create, edit, transfer, lock, and unlock. nc-ext isolates specialized scenarios, including nc-ext-account for deactivating or merging accounts, nc-ext-territory for picking up public-pool leads or accounts and returning them to the public pool, nc-ext-opportunity for finding overdue opportunities, and nc-ext-visit-prepare for visit preparation from account, opportunity, or lead entries. nc-skillbench provides a workspace for generating custom skills around neocrm commands.

Boundaries and cautions

This pack suits agent workflows that already use NeoCRM or neocrm command conventions. It is not a standalone CRM UI, and entity fields or permissions may vary. Before write actions, confirm the target entity exists, the user has permission, and the selected extension skill matches the business scenario.