Governance Core

Evidence Chain

Plan → Confirm → Trace: The governance triad that transforms agent actions into auditable, accountable records.

Definition

What is the Evidence Chain?

The Evidence Chain is MPLP's answer to the fundamental governance question: How do we demonstrate what an agent intended, what was authorized, and what actually happened?

In traditional software, audit trails are often implemented as an afterthought. In agent systems, they must be structural. The Evidence Chain supports that every agent action is preceded by explicit intent (Plan), authorized by governance boundaries (Confirm), and recorded with schema-defined semantics (Trace).

Governance Triad

Plan · Confirm · Trace

Each pillar serves a distinct governance function. Together, they form a continuous chain of evidence.

P

Plan

Intent Declaration

  • Declares what the agent intends to do
  • Specifies why (intent rationale)
  • Defines scope and constraints

Governance Role

Pre-execution intent capture

C

Confirm

Authorization Gate

  • Validates against governance policies
  • Checks permission boundaries
  • Records authorization decision

Governance Role

Execution authorization

T

Trace

Structured Record

  • Records what happened (actions, decisions)
  • Captures state transitions
  • Supports review and reconstruction

Governance Role

Post-execution evidence

Audit Readiness

Why Evidence Chains Matter

In regulated environments, the ability to demonstrate intent, authorization, and execution is often required.

Regulatory Assurance

Auditors often require evidence that agent actions were authorized and within policy boundaries. The Evidence Chain provides structured evidence to support verification.

Incident Investigation

When something goes wrong, the Evidence Chain enables precise reconstruction of the decision path that led to the outcome.

Accountability

Each link in the chain is attributed: who (or what) made the decision, under what authority, and with what constraints.

Trust Foundation

Organizations can deploy agent systems with confidence, knowing that governance is operationalized through explicit lifecycle gates and records.

External Resources

Validation Lab

Evidence-based verdict viewing and export. Coverage reflects evidence maturity — not endorsement, ranking, or certification.

Live counts and verdict details are shown in the Lab.

Positioning Notice: This website provides discovery and positioning content only. For formal protocol definitions, see docs.mplp.io. Source of truth: GitHub Repository.