Codimir AI Fabrick
Differential AGI fabric of role-specialized agents with orchestration, QA, continuations, and MCP tool bindings.
Codimir AI Fabrick
Codimir AI Fabrick is a differential AGI fabric: a system of role-specialized agents that coordinate like a company. A single prompt can activate multiple agents — each with custom instructions, rules, and responsibilities — who assign tasks to each other.
- Roles: 200+ templates imitating real-world jobs.
- Orchestration: overseer/router/proctor pipeline for routing, QA, and escalation.
- MCP Integration: plug-and-play tools (browser, security scan, analytics, docs).
- Continuations: yield/resume across serverless limits.
- Multi-tier: light agents (Next.js) ↔ heavy agents (RunPod/Docker).
- Vision: from projects → verbal → BMI thought-to-task execution.
Sections
- Overview: What is the Fabrick?
- Architecture: Layers and flows
- Roles: Template schema and examples
- Orchestration: Overseer, Router, Proctor
- Standard Flow: Light ↔ Heavy workflow
- MCP: Model Context Protocol bindings
- Roadmap: Phases and milestones
- Playbooks: Continuations, Escalation, QA/Proctor, RunPod Offload
- Concepts: Differential AGI, Fabrick as a Company, MCP as OS, Thought to Task, BMI Future