Everything you need to evaluate, deploy, and champion GuardSpine inside your organization.
Full technical specification covering the 24-department guard model, rubric schema, evidence lifecycle, and integration points.
Download spec (v0.9) →Core engine: guard evaluation loop, rubric compiler, evidence collector, and department orchestrator. Apache 2.0 licensed.
View on GitHub →Command-line tool to run guards against your codebase, policies, or documents. Outputs SARIF-compatible reports.
pip install guardspine-verify Reference YAML schema for defining department rubrics, risk thresholds, escalation paths, and guard type mappings.
View schema →Dogfood evidence packs generated against the GuardSpine kernel itself. Sealed SHA-256 hash chains, rubric audits, test results, and 3-model local council votes.
Sample board-ready AI governance packet generated by GuardSpine. Includes risk heatmap, control status dashboard, and evidence summary.
| Finding | Sev | Owner | Status |
|---|---|---|---|
| Model validation gap in prod LLM | High | ML Eng | In Progress |
| Legal review backlog (3 models) | Med | Legal | In Progress |
| Ethics board charter not renewed | Med | CISO | Open |
Estimate compliance cost reduction, audit prep time savings, and risk exposure reduction with GuardSpine deployment.
2026 regulatory overview: EU AI Act enforcement timeline, NIST AI RMF mapping, and how GuardSpine aligns to each framework.
GuardSpine's 24-department model maps to ISO 42001 AIMS clauses 4-10, providing continuous evidence collection for certification audits.
Enterprise deployment architecture: source integrations, adapter layer, kernel processing, and UI delivery.
Not sure where to start?
Take the Assessment