ToolProduction Ready

datafoundry

Name: datafoundry
Rating: 3.1 (773 reviews)
Author: datagallery-ai

by datagallery-ai

Workspace for running and governing data-focused agents

TypeScript

Updated Sep 22, 2026

773

Stars

110

Forks

View on GitHub

What It Does

Provides an open-source AI workbench that unifies data sources, knowledge, tools, and an agent runtime into a governed workspace for interactive analytics. Exposes an ag-ui and agent runtime so analysts and developers can spin up agents that query data, run transforms, and iterate on results within the same environment. Notable features include integrated data connectors, a workspace UI for exploration, and tooling to compose agent workflows for data analysis. For interoperability, see the AG-UI Protocol and the Open Agent Specification (Agent Spec).

Key Benefits

As agents are given access to sensitive datasets and delegated data tasks, tracing what happened and who made decisions becomes essential for trust. DataFoundry centralizes data access, agent runs, and workspace governance so teams can audit agent behavior and compare outcomes across runs. That clarity helps teams instrument agent track records and evaluate agent reliability in real data workflows rather than synthetic benchmarks. This can be reinforced by applying the Mutual Verification Pattern.

Best For

Data teams and ML engineers who need a governed, interactive environment to run agents on real datasets and audit their behavior.

Applications

Run agents that query and analyze production datasets with governed access controls
Iterate on agent workflows in an interactive workspace and record runs for auditing
Compare agent outputs over time to build an empirical track record for reliability

Avoid common pitfalls

Learn what failures to watch for