Skip to content

Brain Factory

A hub that gives every project an upgradeable, runtime-agnostic brain for AI-assisted delivery — a portable core layer, a project-owned extension layer, and a two-way improvement loop.

Get started See the architecture

Start here

  •  How it works


    A five-minute, plain-language tour of the hub, the brains it provisions, the core/extension split, and the improvement loop.

    Read the overview

  •  Design overview


    The whole target architecture on one page — the moving parts, the diagrams, and what's built vs. planned.

    See the design

  •  Bootstrap a setup


    Go from a natural-language description of your needs to a ready-to-work setup, then verify readiness.

    Prompt-to-setup runbook

Explore the framework

  •  Operating model


    How the framework runs day-to-day: agents and automation, handoffs, issue intake, and routing.

    Operating model

  •  Adoption & profiles


    Reuse the framework across repos and teams — maturity model, profile packs, setup intent, and transplant checklists.

    Adoption & portability

  •  Continuity & memory


    What the framework remembers across sessions: continuity snapshots, artifact indexing, and queued execution.

    Continuity & memory

  •  Governance & policy


    Change governance, release and versioning semantics, security and secure delivery, and the governance checklist.

    Governance & policy

  •  Runbooks


    Step-by-step operator procedures — apply setup, resume from a handoff, triage a Dependabot PR, run health checks.

    All runbooks

  •  Architecture decisions


    The ADR log — every significant design decision, its context, and the trade-offs behind it.

    ADR index

New here?

First-time maintainer or operator? The fastest path:

  1. Read the contract. AGENTS.md at the repository root is the minimum operating contract for agents and contributors.
  2. Get oriented. The Operator onboarding pack covers your first day and first week.
  3. Set up. Describe your needs and run the prompt-to-setup bootstrap, then verify with Apply setup.

For everything else, use the tabs above — each section has its own index. Start with the Operating model, and treat the Issue taxonomy as canonical for issue-template selection.