AI System
A goal-aware assistant built for Copilot CLI that tracks your quarterly objectives, monitors GitLab activity across 30+ OSDU services, and helps you decide what matters. When work shifts from planning to execution, you dispatch a specialist for infrastructure or service operations.
What the System Adds
Daily briefings -- goal progress, open MRs, pipeline failures, and priority recommendations.
Dependency scanning -- risk-scored vulnerability analysis with tiered remediation.
Infrastructure verification -- evidence-based health checks, safeguards validation, systematic debugging.
One-command shipping -- review, quality checks, commit, push, and merge request in a single flow.
Persistent knowledge -- decisions, reports, and goals survive across conversations.
The Core Mental Model
You talk to one assistant. It observes, plans, and ships. When you need to operate on infrastructure or services, you invoke a specialist.
The default assistant knows why (goals, priorities, decisions). Specialists know how (deep domain execution).
| You want to... | Who handles it |
|---|---|
| Observe, plan, decide, ship | The default assistant |
| Modify infrastructure (Terraform, Helm, AKS) | @cimpl specialist |
| Operate on OSDU services (clone, scan, test) | @osdu specialist |
Delegation is one-way: the default dispatches to specialists, never the reverse. You can also invoke a specialist directly without going through the default.
What this looks like in practice
/gm generates a briefing with goal progress and GitLab activity. You review priorities, then dispatch work: osdu, clone partition and scan deps for service work, or cimpl, debug the deployment failure for infrastructure. When done, /send ships your changes.
Developer Personas
| Persona | Role | Primary Experience |
|---|---|---|
| Service Developer | Uses CIMPL as their deployment platform | Default + @osdu (CIMPL fades to background) |
| Platform Engineer | Builds and maintains CIMPL itself | Default + @cimpl + @osdu |
Most developers are service developers. They work across OSDU services and rarely need direct infrastructure control. The @cimpl specialist is primarily for platform engineers making structural changes such as adding middleware, migrating storage backends, or redesigning deployment layers.
Architecture
The system is built in five layers, from end-to-end patterns down to infrastructure:
| Layer | What it provides |
|---|---|
| Workflows | End-to-end patterns: daily briefing, ship-it, debug+verify, clone+scan |
| Agents | @cimpl and @osdu specialists, plus sub-agents for builds and QA |
| Capabilities | Skills and extensions: the shared building blocks |
| Knowledge | Brain (vault for decisions and reports), memory (preferences), and repository context |
| Infrastructure | Extensions for scheduling and live platform access, plus MCP-backed vault search |
The plugin packages agent definitions, shared skills, extensions, and MCP integrations into a portable structure that Copilot CLI loads automatically.
-
Install the plugin, configure your environment, and run your first session.
-
The default assistant,
@cimplfor infrastructure, and@osdufor service operations. -
Skills, extensions, and MCP servers: the building blocks behind every workflow.
-
Persistent knowledge vault with hybrid search, decay rules, and session digests.
-
Repeatable operating patterns: briefing, clone+scan, debug+verify, ship.
Looking for the platform itself?
This plugin provides the AI system for operating the OSDU platform. The platform infrastructure, deployment model, and service architecture are part of the cimpl-azure-provisioning repository.