Glossary¶
Key terms and definitions in the Agentic Cognitive Governance Protocol.
A¶
ACL Tier (Access Control Level Tier)
Defines the strictness of governance applied to an agent. Ranges from ACL-0 (minimal) to ACL-5 (maximum oversight).
Agent
An AI system that makes decisions and takes actions autonomously.
ARS (Agent Risk Score)
Numerical score (0-15) measuring agent autonomy, adaptability, and continuity. Calculated from risk assessment using three dimensions (each 0-5). Used to determine appropriate ACL tier.
ARI Framework
Agent-Reflection-Intervention: The core governance model of ACGP. Describes the flow where agents create traces, governance stewards reflect on them, and appropriate interventions are issued.
Audit Log
Immutable record of all governance events and decisions.
B¶
Blueprint
Configuration defining governance policies, evaluation criteria, and intervention rules for specific use cases.
Block
Intervention level that prevents a specific action from executing.
C¶
Cognitive Trace
Structured record of an agent's reasoning and planned action, submitted for governance evaluation.
Cognitive Trace Quality (CTQ)
Score (0.0-1.0) measuring the quality of an agent's reasoning and planned actions across five standard metrics: reasoning quality, knowledge grounding, ethical alignment, tool safety, and context awareness.
Conformance Level
Implementation level: Minimal, Standard, or Complete, defining which ACGP features are implemented.
E¶
Escalate
Intervention level requesting human approval before proceeding.
Evaluation
Process of assessing cognitive trace quality and determining appropriate intervention.
F¶
Flag
Intervention level that logs action for review but allows it to proceed.
G¶
Governance Steward
Component responsible for evaluating cognitive traces and issuing interventions.
H¶
Halt
Highest intervention level, stopping the agent completely due to critical violations.
I¶
Intervention
Proportionate response to cognitive trace evaluation. Six intervention types (five primary levels: OK, Nudge, Escalate, Block, Halt, plus orthogonal Flag that can be combined with any primary level).
N¶
Nudge
Intervention level suggesting reconsideration but allowing action to proceed.
O¶
OK
Intervention level indicating action can proceed without concerns.
T¶
Tripwire
Hard limit that triggers immediate intervention when exceeded, regardless of reasoning quality.
Trust Debt
Accumulated risk score from flagged actions (0.0+). Accumulates when actions are flagged for review. Decays exponentially by 5% every 24 hours (multiplied by 0.95 daily). When trust debt exceeds thresholds, it triggers re-tiering to higher ACL tiers for stricter oversight.
Trust Score
Overall agent reliability metric (0.0-1.0) tracking behavior over time, where higher is better. Influences governance strictness. Note: Trust Score and Trust Debt are related but distinct concepts - Trust Score represents overall reliability, while Trust Debt specifically tracks flagged behaviors.
See ACGP-1001: Terminology for complete definitions