RIGOR Framework
Research, Inspect, Generate, Optimize, Review - The 5-phase execution framework governing all autonomous agent operations.
The Five Phases
Phase 1: Research
RGather context, validate assumptions, and identify constraints before taking action.
Activities
- Analyze codebase structure and patterns
- Retrieve relevant knowledge from graph
- Identify dependencies and constraints
- Validate initial assumptions
- Build execution context
Outputs
Phase 2: Inspect
IAudit safety, verify constraints, and run pre-execution quality gates.
Activities
- Run security audit checks
- Verify preconditions met
- Execute multi-dimensional quality gates
- Check reversibility requirements
- Validate bounded autonomy
Outputs
Phase 3: Generate
GExecute the task, produce artifacts with confidence scoring and decision documentation.
Activities
- Execute primary operation
- Generate artifacts (code, config, etc.)
- Calculate confidence scores
- Document decisions and reasoning
- Select appropriate patterns
Outputs
Phase 4: Optimize
OTune performance, minimize costs, and trigger learning from the execution.
Activities
- Performance optimization
- Cost minimization analysis
- Feedback integration
- Pattern effectiveness update
- Learning trigger
Outputs
Phase 5: Review
RValidate outcomes, run A/B tests, prepare rollback, and record success metrics.
Activities
- Validate execution outcome
- Run hypothesis testing
- Prepare rollback plan
- Record success metrics
- Update knowledge graph
Outputs
Quality Gates (21 Total)
Pre-deployment validation gates that ensure operational quality and compliance.
v1.5 Gates (Operational Quality)
RIGOR Confidence
Minimum confidence score for execution
Security Scan
Security vulnerability detection
Test Coverage
Minimum code test coverage
Code Review
Automated code review checks
ACMF Validation
Maturity level compliance
Performance Check
Performance baseline comparison
Compliance Check
Regulatory compliance validation
Dependency Audit
Dependency vulnerability scan
Schema Consistency
SQL query schema validation
Docker Build Compliance
Build configuration validation
Image Tag Consistency
Container image verification
v1.6 Gates (Enterprise Governance)
Runtime Confidence
Real-time confidence monitoring
Context Quality
Context pollution detection
Test Coverage Minimum
Enforced test coverage
Audit Trail Present
EU AI Act compliance
Model Drift
Output distribution drift detection
Cost Anomaly
Cost spike prevention
Rollback Capability
Rollback mechanism verification
Human Approval
High-risk operation approval
Supply Chain Integrity
Dependency security
Cascade Protection
Circuit breaker verification
ACMF Integration
RIGOR works in conjunction with the ACMF (Agent Capability Maturity Framework) to control execution flow based on agent maturity level.
Execution Flow with ACMF
RESEARCH → INSPECT → GENERATE → [ACMF Check] → OPERATIONALIZE → REVIEW
↓
┌─────────────────────────┐
│ OBSERVE/LEARNING Mode │
│ ↓ │
│ Status: awaiting_approval
│ (Pause until human approves)
└─────────────────────────┘
↓
┌─────────────────────────┐
│ ASSISTED Mode │
│ ↓ │
│ Critical ops: await approval
│ Normal ops: continue
└─────────────────────────┘
↓
┌─────────────────────────┐
│ AUTONOMOUS Mode │
│ ↓ │
│ Continue execution
│ (Audit trail only)
└─────────────────────────┘Runtime Governance (ACMF)
- • Controls execution during RIGOR phases
- • Based on agent maturity level
- • Automatic progression/regression
- • Per-agent state persistence
Pre-Deploy Governance (Gates)
- • Validates before deployment starts
- • Based on operation risk level
- • Cannot bypass for high-risk ops
- • 24-hour approval timeout
Implementation Example
RIGOR Executor (Go)
// ExecuteWithRIGOR runs an operation through all 5 RIGOR phases
func (re *RIGORExecutor) ExecuteWithRIGOR(
ctx context.Context,
agentID string,
capability string,
acmfMode ACMFMode,
input map[string]interface{},
phases *RIGORPhaseHandlers,
) (*RIGORExecution, error) {
execution := &RIGORExecution{
ID: generateExecutionID(agentID),
AgentID: agentID,
Capability: capability,
ACMFMode: acmfMode,
CurrentPhase: PhaseResearch,
Status: "running",
}
// Phase 1: RESEARCH
researchResult := re.executePhase(ctx, execution, PhaseResearch, phases.Research)
if !researchResult.Success {
return execution, fmt.Errorf("research phase failed")
}
// Phase 2: INSPECT (Quality Gates)
inspectResult := re.executePhase(ctx, execution, PhaseInspect, phases.Inspect)
if !inspectResult.Success {
return execution, fmt.Errorf("inspect phase failed")
}
// Phase 3: GENERATE
generateResult := re.executePhase(ctx, execution, PhaseGenerate, phases.Generate)
// ACMF Check - Human approval for OBSERVE/LEARNING modes
if acmfMode == ACMFModeShadow || acmfMode == ACMFModeLearning {
execution.Status = "awaiting_approval"
return execution, nil // Pause until approved
}
// Phase 4: OPERATIONALIZE
operationalizeResult := re.executePhase(ctx, execution, PhaseOperationalize, phases.Operationalize)
// Phase 5: REVIEW
reviewResult := re.executePhase(ctx, execution, PhaseReview, phases.Review)
execution.Status = "completed"
return execution, nil
}Continue Exploring
Learn more about related architectural components