Governance & Safety

Tool: policy-govern Model: Advanced Requires: ALLOW_GOVERNANCE_SKILLS=true

Trigger & Intent

Triggered by: Compliance checks, prompt injection hardening, and safety validation for regulated workflows.

Intent: Strictly validates outputs against policy controls before allowing interaction with sensitive systems.

Resource Pooling

Capability profile: governance — requires security_audit + adversarial, prefers deep_reasoning, human-in-the-loop required.

Required Skills

Skill	Role
`gov-data-guardrails`	PII detection and data handling compliance
`gov-model-compatibility`	Model capability/policy compatibility check
`gov-model-governance`	Model lifecycle governance
`gov-policy-validation`	Policy schema validation
`gov-prompt-injection-hardening`	Injection attack prevention
`gov-regulated-workflow-design`	Regulated environment workflow design
`gov-workflow-compliance`	End-to-end workflow compliance audit

Input Schema

{
  targetPipeline: string;
  policySchema: string;
}

Decisions & Throw-Backs

If any PII, injection vulnerability, or policy violation is flagged by the adversarial tier → throws back to design or implement loudly. Does not soft-fail silently.

Success Chains

On successful completion chains to: review · resilience · document

FSM — Multi-level governance of action

stateDiagram-v2
    [*] --> PolicyTrigger
    PolicyTrigger --> InjectionDefenseLayer
    PolicyTrigger --> PolicyValidationLayer
    PolicyTrigger --> ComplianceAuditLayer

    InjectionDefenseLayer --> ReactiveBlock
    PolicyValidationLayer --> PolicyOptionSelection
    ComplianceAuditLayer --> LegitimacyCheck

    ReactiveBlock --> GovernanceArbitration
    PolicyOptionSelection --> GovernanceArbitration
    LegitimacyCheck --> GovernanceArbitration

    GovernanceArbitration --> ClearedForExecution: all layers aligned
    GovernanceArbitration --> Inhibit: policy conflict unresolved
    Inhibit --> Reconsideration
    Reconsideration --> PolicyValidationLayer

    ClearedForExecution --> PolicyOutcome
    PolicyOutcome --> ComplianceAudit
    ComplianceAudit --> ComplianceAuditLayer
    ComplianceAudit --> [*]

Execution Sequence

sequenceDiagram
    participant Orchestrator
    participant Pool (Analytical)
    participant Pool (Mechanical)
    participant Tool (Context)

    Orchestrator->>Pool (Analytical): Allocate Capability Profile
    activate Pool (Analytical)
    Pool (Analytical)->>Tool (Context): Issue Tool Calls (Parallel)
    Tool (Context)-->>Pool (Analytical): Return Data

    alt Shallow Loop
        Pool (Analytical)->>Pool (Analytical): Auto-correct Schema
    else Medium Loop
        Pool (Analytical)->>Pool (Mechanical): Delegate Fixes
    end

    Pool (Analytical)-->>Orchestrator: Synthesis Gate
    deactivate Pool (Analytical)

    opt Deep Loop
        Orchestrator->>Orchestrator: Complete Throw-back to Prior Stage
    end