rigor

Epistemic guardrails, knowledge graph, and
metacognition for AI.

Ground every claim. Enforce every constraint. Audit every decision. 

$ brew tap Rigor-Cloud/rigor && brew install rigor

How rigor works

The three pillars

Guardrails

Declarative beliefs and defeaters block hallucinations before they ship.

Knowledge graph

Every claim grounded against your project's facts, stored in a queryable graph.

Metacognition

Rigor reasons about its own confidence -- Gettier guards, induction tracking, credibility decay.

How it works

  1. 01
    Claim extraction

    Rigor parses LLM output into atomic claims with confidence scores.

  2. 02
    Knowledge graph

    Claims are grounded against facts derived from your codebase and constraints.

  3. 03
    Constraint eval

    DF-QuAD argumentation weighs beliefs, justifications, and defeaters.

  4. 04
    Metacognitive audit

    Rigor flags claims where its own confidence is too low to trust.

  5. 05
    Decision

    Allow, warn, or block -- with a receipt of every input that shaped the decision.

Try it today

$ brew tap Rigor-Cloud/rigor && brew install rigor
$ rigor validate rigor.yaml

Requires macOS (Apple Silicon or Intel). Linux support coming.

Cloud, coming soon

Team dashboards, shared constraint graphs, and more.