GenCMS is iGears' model-agnostic AI workspace and runtime direction for building assistants, document extraction, content and internal agent solutions on an organisation's own data.

It is for developers and organisations that need custom or privately deployed AI with explicit tenant, data and permission boundaries.

Account and tenant administration, workspace records, scoped API keys and Guard scanning are available. Composer can run through a tenant's region-locked AIHK or AICA gateway after its AI key is ready.

Request managed workspace access or ask iGears to scope a 30-day MVP. Project services are fixed-price, with the scope, fee and acceptance checks stated in a quotation.

Documentation

Everything, explained once.

20 pages covering connected tools and the product roadmap — from public Guard scanning to the proposed Skill package and workflow architecture.

Start here Preview API reference

Documentation includes product-preview material. Authentication, regional AI gateways, Composer runs and public Guard scans are connected; the public SDK/CLI, unified API, marketplace and automation backends are not.

Get started

The shortest path from an account to a working run.

Getting started

Core concepts

Models, Memory, Skills, Agents, Workflows, Apps, API and Guard — and how they fit together.

Skills

Package format, manifest, execution modes, publishing and versions.

Skills

What a Skill is, how installing works, and what happens on a run.

Skill package structure

The folder layout GenCMS expects, and the smaller subset the MVP accepts.

Manifest reference

Every field in skill.yaml, what it controls and what validation rejects.

Execution modes

Fixed, recommended, auto, user choice and compare — and when each is right.

Publishing a Skill

From local folder to marketplace listing, and what review actually checks.

Versioning and updates

Semantic versions, pinning, and why business Workflows do not auto-update.

Memory

Collections, sources, retrieval and who is allowed to read what.

Memory

Collections, what goes in them, and how retrieval reaches a run.

Adding sources

Supported source types, indexing states and re-indexing.

Memory access control

Granting collections to Agents, Skills and runs — and the audit trail.

Automate

Agents, Workflows and Apps — the three ways work repeats.

Agents

Goals, schedules, memory grants and run history.

Workflows

Node types, branching, approval steps and what the MVP builder supports.

Apps

Packaging components into something a non-technical user can run.

Security

Guard, the permission model, and what each trust label means.

GenCMS Guard

What a scan checks, how the score is produced, and how to run it in CI.

Permission model

The permission set, what each one allows, and what is not available.

Trust levels

Planned definitions for Official, Reviewed, Community and Unverified marketplace labels.

API

Endpoints, errors, limits, SDKs and the CLI.

API reference — deployed routes

The session-authenticated and public endpoints implemented today.

Errors and limits

Error shape, the codes you will actually hit, and what is rate limited.

SDK and CLI — planned

Planned client tooling that is not shipped today.

Playground preview

Inspect the proposed browser request tester.

SDK and CLI preview

Review the proposed client libraries and command line.

Guard API

Scan a package from CI before it reaches production.