ReddRadar
Agent skill
synthesis
Traceability matrix, quality validation, completion decision. Semantic matching for requirement-deliverable traceability (EFL V4.0). Sub-Orchestrator pattern with convergence detection.
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/data/synthesis
SKILL.md
/synthesis - Traceability and Quality Validation
Version: 4.1.0 | EFL Pattern: P1 + P3 + P5 + P6 Role: Sub-Orchestrator for semantic traceability and completion decision Pipeline: After /collect, before /commit-push-pr or /rsil-plan
1. Purpose
Synthesis Sub-Orchestrator that:
- Delegates analysis to specialized agents (P1)
- Performs semantic requirement-deliverable matching (P3)
- Validates quality (consistency, completeness, coherence)
- Executes Phase 3.5 Review Gate (P5)
- Tracks convergence across iterations (P6)
- Makes COMPLETE or ITERATE decision
2. Invocation
/synthesis # Standard (80% threshold)
/synthesis --strict # Strict mode (95% threshold)
/synthesis --lenient # Lenient mode (60% threshold)
/synthesis --dry-run # Analysis only, no decision
/synthesis --workload <slug> # Explicit workload
3. Execution Flow
/synthesis (Sub-Orchestrator)
|
+-- Phase 0: Context Loading
| +-- Read requirements from /clarify
| +-- Read collection report from /collect
| +-- Load iteration history
|
+-- Phase 1: Agent Delegation (P1)
| +-- Agent 1: Phase 3-A Semantic Matching
| +-- Agent 2: Phase 3-B Quality Validation
|
+-- Phase 2: Convergence Detection (P6)
| +-- Compare with previous iteration
| +-- Detect improvement stall
|
+-- Phase 3.5: Review Gate (P5)
| +-- Holistic verification
|
+-- Phase 4: Decision
+-- COMPLETE -> /commit-push-pr
+-- ITERATE -> /rsil-plan
4. L1/L2/L3 Output Format
L1 Summary (returned to main context)
taskId: synthesis-{timestamp}
agentType: synthesis
status: success
summary: "Coverage 85%, Decision: COMPLETE"
decision: "COMPLETE"
coverage: 85.0
threshold: 80
criticalIssues: 0
l2Path: .agent/prompts/{slug}/synthesis/synthesis_report.md
requiresL2Read: false
nextActionHint: "/commit-push-pr"
eflMetrics:
agentDelegation: true
internalIterations: 2
reviewGatePassed: true
converged: false
L2 Report Structure
# Synthesis Report
## Summary
| Metric | Value |
|--------|-------|
| Coverage | 85.0% |
| Threshold | 80% |
| Decision | COMPLETE |
## Traceability Matrix
| Requirement | Status | Deliverables |
|-------------|--------|--------------|
| REQ-001 | covered | file1.py |
## Quality Validation
- Consistency: PASSED
- Completeness: PASSED
- Coherence: PASSED
## Decision
**COMPLETE** - Proceed to /commit-push-pr
5. Semantic Matching (Phase 3-A)
Unlike keyword matching, semantic analysis:
- Understands requirement intent
- Matches conceptual relationships
- Assigns confidence scores (0.0-1.0)
- Considers architectural patterns
Coverage Classification:
- Covered (100%): confidence >= 0.7
- Partial (50%): 0.4 <= confidence < 0.7
- Missing (0%): confidence < 0.4
6. Quality Validation (Phase 3-B)
3C Checks:
- Consistency: No duplicate/conflicting implementations
- Completeness: All P0 requirements covered, tests exist
- Coherence: Components integrate properly
7. Convergence Detection (P6)
Tracks iteration progress:
- Improvement rate < 5% -> converged
- Critical issues not reducing -> escalate
- Coverage plateau (3 iterations < 2%) -> manual review
- Max iterations (5) -> force escalate
8. Decision Logic
| Condition | Decision | Next Action |
|---|---|---|
| coverage >= threshold, critical=0 | COMPLETE | /commit-push-pr |
| coverage >= threshold-20, critical=0 | COMPLETE_WITH_WARNINGS | /commit-push-pr --with-warnings |
| otherwise | ITERATE | /rsil-plan |
9. Integration Points
Input Dependencies
| Source | Path | Purpose |
|---|---|---|
| /clarify | .agent/prompts/{slug}/clarify.yaml |
Requirements |
| /collect | .agent/prompts/{slug}/collection_report.md |
Deliverables |
Output Destinations
| Destination | Path | Purpose |
|---|---|---|
| /commit-push-pr | synthesis_report.md | Completion evidence |
| /rsil-plan | Gaps list | Remediation input |
10. Handoff Contract
handoff:
skill: "synthesis"
workload_slug: "{slug}"
status: "completed"
next_action:
skill: "/commit-push-pr" # or "/rsil-plan"
arguments: "--workload {slug}"
reason: "Coverage {X}% - {COMPLETE|ITERATE}"
Version History
| Version | Change |
|---|---|
| 4.1.0 | Cleaned duplicate blocks, normalized frontmatter |
| 4.0.0 | EFL V4.0 integration (P1, P3, P5, P6) |
| 3.0.0 | Semantic matching, convergence detection |
| 2.2.0 | /rsil-plan integration |
| 1.0.0 | Initial traceability matrix |
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
majiayu000/claude-skill-registry
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
majiayu000/claude-skill-registry
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
majiayu000/claude-skill-registry
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
majiayu000/claude-skill-registry
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
majiayu000/claude-skill-registry
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
majiayu000/claude-skill-registry
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
Didn't find tool you were looking for?