AI & Agents (MCP)

Built for agents, not just "AI-enabled"

Building an application today without a real entry point for AI agents is no longer a serious option. NeoSaaS exposes a Model Context Protocol (MCP) server — the open standard donated to the Linux Foundation by Anthropic — so an agent (Cursor, an internal automation, or any MCP-compatible client) can understand what your application can do and safely make changes to it.

This is not a chatbot bolted onto the admin panel

The MCP endpoint exposes 12 tools across three tiers: read tools to discover what exists (available dashboard modules, content block types), propose tools that turn a natural-language intent into a structured page, article, or dashboard module, and write tools — gated behind a stricter authorization check — that actually publish a change.

Nothing reaches production without review

Every write goes through the same three-step pipeline a human editor would use: validate the proposed change against the schema, preview exactly what will change, then explicitly publish. An agent can draft ten pages in a batch; none of them go live until the publish step runs.

Authentication built for machines and humans alike

An agent authenticates either with a dedicated bearer token or with the same admin session a human uses — both resolve to the same permission model, so an agent never has more access than an admin would.

A full audit trail

Every agent-driven action is logged — what was proposed, what was validated, what was published, and by which actor. Nothing an agent does is invisible after the fact.

Your own AI, your own keys

Beyond the MCP layer, users and admins can store their own LLM provider API keys, encrypted at rest, with per-key usage tracking. NeoSaaS doesn't lock you into a single AI provider.