Skip to main content

AI/ML / Multi Agent Refarch / Controls / DEV

Agent Decision Audit and Explainability

CCC.MARefArc.CN23 · DET

Record an auditable trace of agent decisions, including tool selections, inputs, and rationale, sufficient to explain and review autonomous actions after the fact.

Related Capabilities

IDTitleDescription
CCC.MARefArc.CP19MCP-interaction zero-trust guardrailsEnforces authentication and authorization for every MCP request and governs which agents may use which tools, applying rate limits and validating tool-call parameters.
CCC.MARefArc.CP07Unified sandboxed agent runtimeSecure, sandboxed environment where all agentic reasoning and execution occurs, providing task state management for pause/resume/handoff and intercepting and validating tool calls with credentials handled securely within the sandbox.
CCC.MARefArc.CP08Built-in trusted toolsA collection of bundled, trusted tools providing fundamental capabilities: the MCP client bridge to the external MCP layer, a sandboxed shell, workspace I/O, and web search.
CCC.MARefArc.CP21Human supervision and oversightMechanisms for human reviewers to inspect, approve, correct, or override agent outputs, supporting human-in-the-loop and human-over-the-loop workflows for sensitive or high-impact tasks.
CCC.MARefArc.CP05Agent-ingress zero-trust guardrailsTreats all inputs as untrusted and enforces authentication, authorization, input validation, content filtering, access control, rate limits, and dynamic policy before any request reaches an agent.
CCC.MARefArc.CP02Human-in-the-loop output reviewApplication-embedded controls that allow users to review, approve, or modify agent outputs before they are executed or shared.

Related Threats

IDTitleDescription
CCC.MARefArc.TH27Authorization bypass and tool-chain privilege escalationAgents discover and invoke API endpoints outside their use case, chain individually authorized calls into unauthorized outcomes, circumvent segregation-of-duties workflows, or experience permission creep during operation, defeating intended authorization boundaries.
CCC.MARefArc.TH28Tool selection, parameter, and sequencing manipulationCrafted inputs cause agents to select inappropriate tools, inject malicious parameters into legitimate calls, reorder tool execution into dangerous combinations, corrupt tool-state understanding, or pass one tool's output as malicious input to the next.
CCC.MARefArc.TH25Non-compliant outputs and model-risk-management gapsAI-generated advice, marketing, or communications that fail KYC, suitability, disclosure, record-keeping, or model-risk-management expectations create regulatory exposure; weak supervision and accountability lines turn this into direct non-compliance.

Assessment Requirements

IDTextApplicability
CCC.MARefArc.CN23.AR01The system MUST record an auditable decision trace for each agent action, including the tools invoked and the inputs and rationale leading to them.tlp-clear, tlp-green, tlp-amber, tlp-red
CCC.MARefArc.CN23.AR02Decision traces MUST be retained and reviewable to support explainability and post-incident analysis.tlp-clear, tlp-green, tlp-amber, tlp-red

Guideline Mappings

FrameworkIDRemarks
finos-airAIR-DET-021