LIBER ///

Structured decision knowledge for AI agents

The missing layer between agent identity and agent competence. Decision provenance. Knowledge quality certification. Auditable decision receipts. Delivered via MCP.

Core Capabilities
Decision Provenance
Every agent decision traced to its sources, confidence level, considered alternatives, and rationale — in structured, machine-readable, verifiable format.
Knowledge Quality Certification
Source authority, freshness, internal consistency, and completeness — measured, scored, and embedded as metadata that agents consume before they act, not after damage is done.
Decision Quality Scoring
A standardized metric for agent decision competence. Compare agents. Set minimum thresholds. Hold autonomous systems accountable for how well they decide, not just what they access.
In Development
LIBER is an independent research project building structured decision knowledge infrastructure for AI agents. MCP-native. Sub-100ms delivery. Zero LLM on read path. First domain: software engineering decisions.

Standards engagement: NIST NCCoE AI Agent Identity & Authorization. EU AI Act compliance readiness.