> ## Documentation Index
> Fetch the complete documentation index at: https://docs.models.dunsky.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Repository directory

> Dunsky documentation for Repository directory: find model guidance, source references, and the right tools for your work.

Use this directory to distinguish model code, the cloud application, execution services, and assistant tooling. Repository names are implementation references; start with the [model catalogue](https://docs.models.dunsky.com/en/models/catalogue.md) when you need a model’s purpose.

| Repository | Responsibility | Start here |
| - | - | - |
| `packages` | Python modelling packages, including DERP, eFleet, HEAT, EVA, SAM, and LDM | [README](https://github.com/dunsky-consulting/packages/blob/trunk/README.md) |
| `core-app` | Browser application and its application backend | [Local setup](https://docs.models.dunsky.com/en/development/setup.md) |
| `core-api` | Model execution microservice | [README](https://github.com/dunsky-consulting/core-api/blob/main/README.md) |
| `core-mcp` | Reusable assistants exposed through MCP | [README](https://github.com/dunsky-consulting/core-mcp/blob/main/README.md) |
| `platform-infra` | Infrastructure repository referenced by core-app | [Repository](https://github.com/dunsky-consulting/platform-infra) |

## Production and development

Use [models.dunsky.com](https://models.dunsky.com) for normal platform use. Use [dev.models.dunsky.com](https://dev.models.dunsky.com) for development and validation. The development environment is distinct from production; confirm the environment before reviewing a run or reporting a problem.

## Runtime versions vary

The core-app backend requires Python 3.15 or newer. The reviewed model package READMEs use Python 3.14; core-mcp documents Python 3.12. Follow the repository and package you are working on instead of assuming one Python version for everything.

GitHub links require the appropriate repository access.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.