llms.txt

# Predictive-ML-Core

> Scores architecture risk from a declared system inventory (name, type,
> criticality, public exposure, auth method) using explicit, weighted rules —
> not a trained model. Every score ships with a rule-by-rule breakdown
> (`scoreBreakdown`) tying each point back to a specific field in the request
> and a stated rationale, so the reasoning is auditable rather than a black
> box. Deliberately not ML: the request contract carries no runtime
> telemetry and there is no labeled training set of past incidents to fit a
> model against — see the README's "Why this is not a trained model" section.

It is a C#/.NET 10 HTTP service, a CLI, a NuGet package
(`CloudSealed.ML.Core`), and a GitHub Action.

## Docs

- [README](https://github.com/cloudsealed/Predictive-ML-Core/blob/main/README.md): install, HTTP/CLI/GitHub Action usage, alerts, HTML reports, score breakdown example
- [METHODOLOGY.md](https://github.com/cloudsealed/Predictive-ML-Core/blob/main/METHODOLOGY.md): full scoring method
- [ARCHITECTURE.md](https://github.com/cloudsealed/Predictive-ML-Core/blob/main/ARCHITECTURE.md): codebase design ("No I/O, no web" in the core engine)
- [src/CloudSealed.ML.Engine/Scoring/RiskRules.cs](https://github.com/cloudsealed/Predictive-ML-Core/blob/main/src/CloudSealed.ML.Engine/Scoring/RiskRules.cs): every weight, as a named constant with a one-line rationale

## MCP

An MCP server exposing this engine as a `cloudsealed_score_architecture_risk`
tool for AI agents is available at
[cloudsealed/cloudsealed-mcp](https://github.com/cloudsealed/cloudsealed-mcp).

## Optional

- [NuGet package](https://www.nuget.org/packages/CloudSealed.ML.Core): `dotnet add package CloudSealed.ML.Core`
- [Docker image](https://hub.docker.com/r/cloudsealed/predictive-ml-core): `docker pull cloudsealed/predictive-ml-core`
- [GitHub Action listing](https://github.com/marketplace/actions/predictive-ml-core-architecture-risk-audit)