Foundry Dev Tools

MDFactory has a first-class Foundry backend in the sync layer. That backend is implemented through foundry_dev_tools plus the fdt command-line tooling used by the interactive config flow.

What the code expects

The current Foundry path relies on:

• the Python package foundry_dev_tools
• the fdt command being available on PATH

mdfactory sync init config explicitly runs fdt config, then validates connectivity by constructing FoundryContext.

Configure the backend

Set the backend type:

[database]
TYPE = foundry

Then run the interactive wizard:

mdfactory sync init config

What the wizard does:

1. Runs fdt config
2. Checks for Foundry Dev Tools config files in standard locations
3. Validates that FoundryContext() can connect
4. Prompts for a Foundry BASE_PATH
5. Prompts for ANALYSIS_NAME
6. Derives dataset paths for runs, molecules, analyses, and artifacts
7. Writes ~/.mdfactory/config.ini

Derived dataset layout

Given:

[foundry]
BASE_PATH = /Group Functions/mdfactory
ANALYSIS_NAME = analysis

the config flow derives:

• runs: /Group Functions/mdfactory/runs
• molecules: /Group Functions/mdfactory/molecules
• analysis root: /Group Functions/mdfactory/analysis
• artifacts root: /Group Functions/mdfactory/artifacts

Within that layout, the code maps:

• ANALYSIS_OVERVIEW -> <ANALYSIS_DB_PATH>/overview
• ANALYSIS_<NAME> -> <ANALYSIS_DB_PATH>/<name>
• ARTIFACT_<NAME> -> <ARTIFACT_DB_PATH>/<name>

Validate the setup

mdfactory sync init check

When [database] TYPE = foundry, this checks:

• Foundry authentication and connectivity
• the configured base path
• the configured runs, molecules, analysis, and artifact dataset roots

Initialize schemas

After configuration, initialize the datasets used by the sync commands:

mdfactory sync init systems
mdfactory sync init analysis
mdfactory sync init artifacts

This is the recommended way to establish the expected tables or datasets before pushing data.

Push and pull with Foundry

Push run metadata:

mdfactory sync push systems build_summary.yaml

Push analyses:

mdfactory sync push analysis build_summary.yaml

Push artifacts:

mdfactory sync push artifacts build_summary.yaml

Query overview data:

mdfactory sync pull analysis --overview
mdfactory sync pull artifacts --overview

Backend behavior

The Foundry backend is implemented by FoundryDataSource, which uses FoundryContext and dataset paths from the active config. MDFactory still uses the same high-level sync commands regardless of backend; only the storage implementation changes.

Next steps