Built to plug into your toolchain

The MVP runs entirely on local data. This is the architecture it's designed to grow into — each integration slots in behind a seam that already exists in the code.

Nothing here is live yet. These cards describe planned capabilities. The AI panels already read through a getAIInsight() seam, so swapping authored content for a real model or MCP server is a change in one file.

Planned

AWS MCP Server

A Model Context Protocol server that keeps templates honest against live AWS reality.

Read the latest AWS documentation on demand
Review generated Terraform for drift and bad defaults
Explain AWS error messages in plain language
Suggest cost-aware instance and storage choices

Planned

GitHub

Turn a template into a real repository and pull request, not a copy-paste.

Scaffold a repo from any template
Open a PR with the generated workflow and IaC
Compare two maturity levels as a diff
Surface Actions run status inline

Live

DeepSeek API

The AI panels call DeepSeek live (OpenAI-compatible) behind the getAIInsight() seam, with authored content as a graceful fallback.

Explain architecture, Terraform, and pipelines
Review a template's security posture
Generate READMEs and runbooks
Falls back to authored content when no key is set

Planned

Observability Stack

Wire CloudWatch, Prometheus, and Grafana into the higher maturity levels.

Ship logs and metrics by default
Bundle starter dashboards
Attach basic alerting and a runbook