PlatformsComing soon

MANEF AI

Compose the reviewed MANEF AI modules — Models, MCP Builder, Connectors and Agent@Manef — without merging their codebases.

For Teams wiring models, MCP servers and connectors into one product foundation.

Open component setup guides Request repository access

MANEF AI

What you can explore

One descriptor per module

Each product keeps its own repository, runtime, lockfile and credentials; the workspace only adds pinned metadata.

Check before adopting

The offline conformance checker verifies descriptor and lock agreement before any module touches your code.

Preview, do not execute

The local hub shows module readiness and composition previews while reporting adapters as unconfigured.

A task to try

As an integrator, I want one reviewed workspace that pins Models, MCP Builder, Connectors and Agent@Manef so I can adopt a module without forking four codebases.

Your first useful task

Check the requirements, follow the first step, then review the result.

What you need
Node 22.13+ and pnpm 10.34.7 for the workspace. Each module keeps its own runtime: Node for Models, Bun for MCP Builder, a Bash-capable shell for Connectors, Python 3.10+ for Agent@Manef. Access to the private repository is granted by the owner.
Start here
Run pnpm install --frozen-lockfile --ignore-scripts, then pnpm run build and pnpm run check to verify every module descriptor offline before adopting one.
Open component setup guides
What you get
A verified foundation with pinned module revisions, a passing conformance check and a local hub that reports unconfigured adapters instead of pretending to be live.

Availability

Early source toolkit

Private source workspace with pinned module descriptors, an offline conformance checker and a local preview hub. It does not deploy, start products, resolve secrets or route live model calls; per-component setup stays at resource.manef.dev.

Published by: MANEF