Show HN: Interview your analysts instead of auto-generating a semantic layer

Category: ai-ml

Tags: semantic-layer, agent-skills, data-governance, analytics, llm-workflow, analytics-context

Score: 6.5/10 (Innovation: 7, Technical: 5, Documentation: 8, Utility: 6)

Nodal Analytics is a set of agent skills that guides an interactive interview with domain experts to build governed, reviewable analytics context (metric definitions, canonical tables, mandatory filters, join fan-out caveats) rather than auto-generating a semantic layer. It packages the output in a versioned Analytics Context Format (ACF) and includes workflows for planning data questions, verifying answers, and seeding evals. The project is interesting for its human-in-the-loop governance stance and its no-telemetry, local-first design that keeps warehouse credentials and data inside the user's own agent environment.

Target audience: analytics engineers, data engineers, domain analysts, and data platform teams working with AI coding agents

Repository: https://github.com/nodal-data/nodal-context · Python · Apache-2.0 · 10 stars

View on Hacker News