Rsil Plan

/rsil-plan - Requirements-Synthesis Integration Loop

Version: 3.1.0 | EFL Pattern: P1 + P2 + P5 + P6 Role: Gap Analysis and Remediation Planning Pipeline: After /synthesis ITERATE, before /orchestrate or /clarify

1. Purpose

RSIL handles iteration cycles when /synthesis returns ITERATE:

1. Verifies gaps at code level (not just keyword matching)
2. Collects evidence from codebase
3. Classifies gaps (COVERED/PARTIAL/MISSING)
4. Creates remediation tasks
5. Decides AUTO_REMEDIATE or ESCALATE

2. Invocation

/rsil-plan                          # Auto-detect iteration
/rsil-plan --iteration 2            # Specific iteration
/rsil-plan --auto-remediate         # Skip user confirmation
/rsil-plan --workload <slug>        # Explicit workload

3. Pipeline Integration

/synthesis
    |
    +-- ITERATE --> /rsil-plan <-- THIS SKILL
                        |
                        +-- Phase 1: Parallel Explore Agents (P2)
                        +-- Phase 2: L1 Aggregation
                        +-- Phase 3-A: L2 Horizontal Synthesis
                        +-- Phase 3-B: L3 Vertical Verification
                        +-- Phase 3.5: Review Gate (P5)
                        +-- Phase 4: Selective Feedback (P4)
                        +-- Phase 5: User Decision
                        |
                        +-- AUTO_REMEDIATE --> /orchestrate
                        +-- ESCALATE --> /clarify

4. L1/L2/L3 Output Format

L1 Summary (returned to main context)

taskId: rsil-{iteration}-{timestamp}
agentType: rsil-plan
status: success
summary: "RSIL complete: 3 gaps, 2 tasks, EFL verified"

priority: HIGH
decision: "AUTO_REMEDIATE"
gapStats:
  total: 3
  missing: 1
  partial: 2
  highComplexity: 0

l2Path: .agent/prompts/{slug}/rsil-plan/iteration_{n}.md
requiresL2Read: false
nextActionHint: "/orchestrate --workload {slug}"

eflMetrics:
  totalAgents: 2
  parallelPhases: 3
  feedbackLoops: 1
  internalIterations: 4

L2 Report Structure

# RSIL Gap Analysis Report - Iteration {n}

## Summary
| Metric | Value |
|--------|-------|
| COVERED | 5 (62%) |
| PARTIAL | 2 (25%) |
| MISSING | 1 (13%) |

## Decision
**AUTO_REMEDIATE** - Proceeding with /orchestrate

## Gap Details

### MISSING
| ID | Requirement | Complexity |
|----|-------------|------------|
| REQ-003 | Error handling | medium |

### PARTIAL
| ID | Requirement | Evidence |
|----|-------------|----------|
| REQ-002 | Auth flow | src/auth.ts (1 match) |

## Remediation Tasks Created
| Task ID | Requirement | Complexity |
|---------|-------------|------------|
| #101 | REQ-003 | medium |

## Next Action
/orchestrate --workload {slug}

5. Gap Classification

6. Evidence Collection

// Generate Grep patterns from requirement
const patterns = [
  { regex: "(function|class|def).*keyword", fileType: "ts,js,py" },
  { regex: "(describe|test|it).*keyword", fileType: "ts,js" },
  { regex: "(import|from).*keyword", fileType: "ts,js,py" }
];

// Search codebase
for (pattern of patterns) {
  results = await Grep({ pattern: pattern.regex, type: pattern.fileType });
  evidence.files.push(...results.files);
}

7. Decision Logic

8. Remediation Task Creation

const task = await TaskCreate({
  subject: `[RSIL-${iteration}] ${action}: ${requirement}`,
  description: generateTaskDescription(gap),
  activeForm: `Remediating: ${requirementId}`
});

9. Input Dependencies

10. Output Destinations

11. Handoff Contract

handoff:
  skill: "rsil-plan"
  workload_slug: "{slug}"
  status: "completed"
  next_action:
    skill: "/orchestrate"  # or "/clarify"
    arguments: "--workload {slug}"
    reason: "Gap analysis complete, {n} tasks created"

12. Configuration

const DEFAULT_CONFIG = {
  gapThreshold: 3,           // Max gaps for auto-remediate
  highComplexityThreshold: 1, // Max high-complexity gaps
  maxIterations: 5,          // Max RSIL iterations
  evidenceThreshold: {
    covered: 3,              // Matches for COVERED
    partial: 1               // Matches for PARTIAL
  }
};

Version History