Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Validates entire engineering proof chain. Verifies architecture, backend maps, backend code, standardization, frontend types, infrastructure topology all compose correctly. This is the final DEPLOYMENT GATE - deployment blocked if proof chain invalid. Use when engineering thread completes all actions.
.claude/skills/microck-proof-composer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 281% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 235% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 65% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 217% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 429% | 0% |
Validate entire engineering proof chain and authorize deployment. This is the final deployment gate - nothing deploys without a valid composed proof.
Key insight: Each skill generates local proofs. proof-composer verifies they compose into a valid system-wide proof.
Process:
You are skill #6 of 6 (FINAL GATE):
All proofs from previous skills:
artifacts/engineering/proofs/
├── architecture/
│ ├── adt-correctness/
│ ├── functor-laws/
│ ├── composition-correctness/
│ ├── state-machines/
│ └── curry-howard-proofs/
│
├── backend/
│ ├── map-validation/
│ │ └── validation-report.json
│ ├── implementation-correctness/
│ ├── standardization/
│ │ └── naturality-certificate.proof
│ └── runtime-verification/
│
├── frontend/
│ └── type-correspondence/
│ └── openapi-to-typescript.proof
│
└── infrastructure/
└── topology/
└── deployment-isomorphism.proofSpecification manifest:
artifacts/engineering/specifications/manifest.jsonGenerated artifacts:
artifacts/engineering/
├── specifications/v{X}/ # From system-architect
├── maps/backend/ # From backend-prover Phase 1
├── code/backend/ # From backend-prover Phase 2
├── code/frontend/ # From frontend-prover
└── configs/ # From infrastructure-proverpythondef collect_proofs(): """ Collect all proof files from engineering artifacts Returns dict of proofs by category """ proofs = { 'architecture': collect_architecture_proofs(), 'backend': collect_backend_proofs(), 'frontend': collect_frontend_proofs(), 'infrastructure': collect_infrastructure_proofs() } return proofs def collect_architecture_proofs(): """Collect from system-architect""" return { 'adt_correctness': load_proof('architecture/adt-correctness/'), 'functor_laws': load_proof('architecture/functor-laws/'), 'composition': load_proof('architecture/composition-correctness/'), 'state_machines': load_proof('architecture/state-machines/'), 'curry_howard': load_proof('architecture/curry-howard-proofs/') } def collect_backend_proofs(): """Collect from backend-prover + standardization-layer""" return { 'map_validation': load_json('backend/map-validation/validation-report.json'), 'implementation': load_proof('backend/implementation-correctness/'), 'standardization': load_json('backend/standardization/naturality-certificate.proof'), 'runtime': load_proof('backend/runtime-verification/') } def collect_frontend_proofs(): """Collect from frontend-prover""" return { 'type_correspondence': load_json('frontend/type-correspondence/openapi-to-typescript.proof') } def collect_infrastructure_proofs(): """Collect from infrastructure-prover""" return { 'topology': load_json('infrastructure/topology/deployment-isomorphism.proof') }
pythondef verify_version_consistency(proofs, manifest): """ Verify all proofs reference same specification version Critical: Prevents race conditions where different skills used different versions of the spec """ errors = [] # Get spec version from manifest spec_version = manifest['version'] spec_hash = manifest['hash'] # Check architecture proofs arch_version = proofs['architecture'].get('specification_version') if arch_version != spec_version: errors.append({ "type": "version_mismatch", "skill": "system-architect", "expected": spec_version, "actual": arch_version }) # Check backend proofs backend_version = proofs['backend']['map_validation'].get('specification_version') if backend_version != spec_version: errors.append({ "type": "version_mismatch", "skill": "backend-prover", "expected": spec_version, "actual": backend_version }) # Check frontend proofs frontend_version = proofs['frontend']['type_correspondence'].get('specification_version') if frontend_version != spec_version: errors.append({ "type": "version_mismatch", "skill": "frontend-prover", "expected": spec_version, "actual": frontend_version }) # Check infrastructure proofs infra_version = proofs['infrastructure']['topology'].get('specification_version') if infra_version != spec_version: errors.append({ "type": "version_mismatch", "skill": "infrastructure-prover", "expected": spec_version, "actual": infra_version }) return { "consistent": len(errors) == 0, "specification_version": spec_version, "specification_hash": spec_hash, "errors": errors }
pythondef validate_proof_chain(proofs): """ Verify proof chain composes correctly: Requirements → Architecture (system-architect) ✓ ↓ Backend Maps (backend-prover Phase 1) ✓ ↓ Backend Code (backend-prover Phase 2) ✓ ↓ Standardization (standardization-layer) ✓ ↓ Frontend (frontend-prover) ✓ ↓ Infrastructure (infrastructure-prover) ✓ """ errors = [] # 1. Architecture proofs valid if not architecture_proofs_valid(proofs['architecture']): errors.append({ "type": "invalid_architecture_proof", "details": "Architecture proofs failed validation" }) # 2. Backend maps validated if proofs['backend']['map_validation']['status'] != 'valid': errors.append({ "type": "invalid_backend_maps", "details": "Backend maps not validated" }) # 3. Backend code matches maps if not backend_code_matches_maps(proofs['backend']['implementation']): errors.append({ "type": "code_map_mismatch", "details": "Backend code doesn't match verified maps" }) # 4. Standardization preserves composition if not standardization_preserves_composition(proofs['backend']['standardization']): errors.append({ "type": "standardization_broken", "details": "Standardization doesn't preserve composition laws" }) # 5. Frontend types correspond to backend API if not proofs['frontend']['type_correspondence']['bijection']: errors.append({ "type": "type_correspondence_failed", "details": "Frontend types don't match backend API" }) # 6. Infrastructure topology matches architecture if not proofs['infrastructure']['topology']['isomorphism']: errors.append({ "type": "topology_isomorphism_failed", "details": "Deployed topology doesn't match architecture" }) return { "chain_valid": len(errors) == 0, "errors": errors }
pythondef check_for_gaps(proofs, artifacts): """ Check for missing proofs or incomplete verification """ gaps = [] # Check all services in architecture have backend maps arch_services = extract_services_from_architecture(proofs['architecture']) backend_maps = list_backend_maps(artifacts['maps/backend']) for service in arch_services: if service not in backend_maps: gaps.append({ "type": "missing_backend_map", "service": service, "fix": "Run backend-prover to generate map" }) # Check all backend maps have implementations for map_file in backend_maps: service = extract_service_name(map_file) if not backend_implementation_exists(service, artifacts['code/backend']): gaps.append({ "type": "missing_implementation", "service": service, "fix": "Run backend-prover Phase 2" }) # Check all backend services have deployment configs backend_services = list_backend_services(artifacts['code/backend']) for service in backend_services: if not deployment_config_exists(service, artifacts['configs']): gaps.append({ "type": "missing_deployment_config", "service": service, "fix": "Run infrastructure-prover" }) # Check frontend consumes all public APIs public_apis = extract_public_apis(proofs['architecture']) frontend_types = list_frontend_types(artifacts['code/frontend']) for api in public_apis: if not frontend_type_exists(api, frontend_types): gaps.append({ "type": "missing_frontend_type", "api": api, "fix": "Run frontend-prover" }) return { "complete": len(gaps) == 0, "gaps": gaps }
pythondef validate_end_to_end_composition(proofs): """ Verify composition laws hold across entire system """ errors = [] # Architecture composition valid if not proofs['architecture']['composition']['valid']: errors.append({ "type": "architecture_composition_invalid", "details": "Architecture-level composition laws violated" }) # Maps preserve architecture composition if not maps_preserve_architecture(proofs['backend']['map_validation']): errors.append({ "type": "maps_dont_preserve_architecture", "details": "Backend maps don't preserve architecture composition" }) # Code preserves map composition if not code_preserves_maps(proofs['backend']['implementation']): errors.append({ "type": "code_doesnt_preserve_maps", "details": "Backend code doesn't preserve map composition" }) # Standardization preserves composition (naturality) if not proofs['backend']['standardization']['composition_preserved']: errors.append({ "type": "standardization_breaks_composition", "details": "Middleware breaks composition laws" }) # Frontend types compose if not frontend_types_compose(proofs['frontend']['type_correspondence']): errors.append({ "type": "frontend_types_dont_compose", "details": "Frontend type definitions don't compose" }) # Infrastructure topology correct if not proofs['infrastructure']['topology']['isomorphism']: errors.append({ "type": "topology_incorrect", "details": "Infrastructure doesn't match architecture" }) return { "composition_valid": len(errors) == 0, "errors": errors }
pythondef generate_composed_proof(validation_results, proofs, manifest): """ Generate final composed proof certificate This certificate authorizes deployment """ all_checks_passed = ( validation_results['version_consistency']['consistent'] and validation_results['proof_chain']['chain_valid'] and validation_results['gaps']['complete'] and validation_results['composition']['composition_valid'] ) certificate = { "status": "valid" if all_checks_passed else "invalid", "timestamp": datetime.utcnow().isoformat() + "Z", "specification": { "version": manifest['version'], "hash": manifest['hash'] }, "composition_chain": [ "architecture → backend_maps", "backend_maps → backend_code", "backend_code → standardization", "backend_code → frontend", "backend_code → infrastructure" ], "verification_summary": { "version_consistency": validation_results['version_consistency']['consistent'], "proof_chain_valid": validation_results['proof_chain']['chain_valid'], "no_gaps": validation_results['gaps']['complete'], "composition_valid": validation_results['composition']['composition_valid'] }, "individual_proofs": { "architecture": { "adt_correct": True, "functors_valid": True, "composition_correct": True, "state_machines_complete": True }, "backend": { "maps_validated": proofs['backend']['map_validation']['status'] == 'valid', "code_correct": True, "standardization_natural": proofs['backend']['standardization']['composition_preserved'] }, "frontend": { "type_correspondence": proofs['frontend']['type_correspondence']['bijection'] }, "infrastructure": { "topology_isomorphism": proofs['infrastructure']['topology']['isomorphism'] } }, "gaps_detected": validation_results['gaps']['gaps'], "deploy_authorized": all_checks_passed, "errors": ( validation_results['version_consistency']['errors'] + validation_results['proof_chain']['errors'] + validation_results['composition']['errors'] ) } return certificate
Success case:
json{ "status": "valid", "timestamp": "2025-01-15T10:30:00Z", "specification": { "version": "v1.2.0", "hash": "sha256:abc123..." }, "composition_chain": [ "architecture → backend_maps", "backend_maps → backend_code", "backend_code → standardization", "backend_code → frontend", "backend_code → infrastructure" ], "verification_summary": { "version_consistency": true, "proof_chain_valid": true, "no_gaps": true, "composition_valid": true }, "individual_proofs": { "architecture": { "adt_correct": true, "functors_valid": true, "composition_correct": true, "state_machines_complete": true }, "backend": { "maps_validated": true, "code_correct": true, "standardization_natural": true }, "frontend": { "type_correspondence": true }, "infrastructure": { "topology_isomorphism": true } }, "gaps_detected": [], "deploy_authorized": true, "errors": [] }
Failure case:
json{ "status": "invalid", "timestamp": "2025-01-15T10:30:00Z", "verification_summary": { "version_consistency": false, "proof_chain_valid": true, "no_gaps": false, "composition_valid": true }, "gaps_detected": [ { "type": "missing_deployment_config", "service": "BillingService", "fix": "Run infrastructure-prover" } ], "deploy_authorized": false, "errors": [ { "type": "version_mismatch", "skill": "frontend-prover", "expected": "v1.2.0", "actual": "v1.1.0" } ] }
artifacts/engineering/proofs/composed/
├── system-proof.certificate
├── composition-graph.dot
└── verification-report.mdBuild pipeline final gate:
bash# build-pipeline/13-compose-proofs.sh ⭐ DEPLOYMENT GATE echo "Validating system-wide proof chain..." # Run proof-composer proof-composer validate-all # Check certificate status=$(jq -r '.status' artifacts/engineering/proofs/composed/system-proof.certificate) deploy_authorized=$(jq -r '.deploy_authorized' artifacts/engineering/proofs/composed/system-proof.certificate) if [ "$status" != "valid" ] || [ "$deploy_authorized" != "true" ]; then echo "❌ DEPLOYMENT BLOCKED: Proof chain invalid" echo "" echo "Errors:" jq '.errors' artifacts/engineering/proofs/composed/system-proof.certificate echo "" echo "Gaps:" jq '.gaps_detected' artifacts/engineering/proofs/composed/system-proof.certificate exit 1 fi echo "✅ Proof chain valid - deployment authorized" # build-pipeline/14-deploy.sh can now proceed
✓ All proofs collected ✓ Version consistency verified ✓ Proof chain validated ✓ No gaps detected ✓ Composition laws hold end-to-end ✓ System proof certificate generated ✓ Deployment authorized ✓
Version mismatch:
ERROR: Version inconsistency detected
Frontend-prover used v1.1.0, but current spec is v1.2.0
Action: Re-run frontend-prover with correct versionBroken proof chain:
ERROR: Proof chain broken
Backend maps not validated (map-validation/validation-report.json status != 'valid')
Action: Re-run backend-prover Phase 1 (map validation)Gap detected:
ERROR: Missing deployment config
Service: BillingService
Has backend implementation but no deployment config
Action: Run infrastructure-proverComposition broken:
ERROR: Composition doesn't hold end-to-end
Standardization broke composition laws (naturality failed)
Action: Re-run standardization-layer with correct naturality proof threads/engineering/{requirement}/5-actions/action-5-proof-composition.md
Status: Complete Proof status: valid Deploy authorized: true Certificate: artifacts/engineering/proofs/composed/system-proof.certificate Specification: v1.2.0 (hash: abc123...) All proofs verified: ✓ No gaps detected: ✓ Composition valid: ✓
System-Architect
↓
[Architecture Proofs: ADT, Functors, Composition, State Machines] ✓
↓
Backend-Prover Phase 1
↓
[Backend Maps Validated: Composition laws verified] ✓
↓
Backend-Prover Phase 2
↓
[Backend Code Matches Maps: Generated from verified maps] ✓
↓
Standardization-Layer
↓
[Naturality Proven: Middleware preserves composition] ✓
↓
Frontend-Prover
↓
[Type Correspondence: Frontend ≅ Backend API] ✓
↓
Infrastructure-Prover
↓
[Topology Isomorphism: Deployed ≅ Architecture] ✓
↓
Proof-Composer (YOU)
↓
[System Proof: All proofs compose ✓]
↓
DEPLOYMENT AUTHORIZED ✅You are the final gatekeeper. Verify the entire proof chain. Authorize deployment only when mathematically verified.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 5,019 | 20,175 | +302% | 1 | 1 | 0% | 239 | 5,688 | +2280% | 0 | 0 | — |
case-02 | pass→fail | 12,186 | 29,305 | +140% | 1 | 1 | 0% | 2,539 | 5,810 | +129% | 0 | 0 | — |
case-03 | fail→fail | 8,675 | 23,016 | +165% | 1 | 1 | 0% | 1,725 | 6,269 | +263% | 0 | 0 | — |
case-04 | fail→pass | 8,300 | 16,868 | +103% | 1 | 1 | 0% | 1,782 | 6,784 | +281% | 0 | 0 | — |
case-05 | fail→pass | 11,285 | 7,863 | -30% | 1 | 1 | 0% | 2,028 | 6,785 | +235% | 0 | 0 | — |
case-06 | fail→fail | 11,515 | 15,859 | +38% | 1 | 1 | 0% | 2,144 | 6,026 | +181% | 0 | 0 | — |
case-07 | fail→fail | 9,787 | 17,410 | +78% | 1 | 1 | 0% | 1,876 | 6,120 | +226% | 0 | 0 | — |
case-08 | fail→fail | 13,737 | 22,003 | +60% | 1 | 1 | 0% | 2,744 | 6,360 | +132% | 0 | 0 | — |
case-09 | fail→fail | 8,687 | 15,332 | +76% | 1 | 1 | 0% | 1,421 | 6,094 | +329% | 0 | 0 | — |
case-15 | fail→pass | 22,406 | 7,497 | -67% | 1 | 1 | 0% | 4,042 | 6,658 | +65% | 0 | 0 | — |
case-10 | fail→pass | 11,165 | 7,228 | -35% | 1 | 1 | 0% | 2,087 | 6,621 | +217% | 0 | 0 | — |
case-11 | fail→fail | 16,900 | 21,527 | +27% | 1 | 1 | 0% | 2,718 | 6,032 | +122% | 0 | 0 | — |
case-12 | fail→pass | 7,770 | 15,162 | +95% | 1 | 1 | 0% | 1,254 | 6,639 | +429% | 0 | 0 | — |
case-13 | fail→pass | 31,196 | 12,147 | -61% | 1 | 1 | 0% | 2,961 | 7,511 | +154% | 0 | 0 | — |
case-14 | fail→pass | 9,542 | 6,585 | -31% | 1 | 1 | 0% | 1,480 | 6,392 | +332% | 0 | 0 | — |
case-16 | fail→pass | 13,998 | 15,299 | +9% | 1 | 1 | 0% | 944 | 6,831 | +624% | 0 | 0 | — |
case-17 | fail→fail | 9,795 | 13,921 | +42% | 1 | 1 | 0% | 1,864 | 5,848 | +214% | 0 | 0 | — |
case-18 | pass→pass | 5,591 | 3,442 | -38% | 1 | 1 | 0% | 849 | 5,899 | +595% | 0 | 0 | — |
case-19 | pass→pass | 7,845 | 3,904 | -50% | 1 | 1 | 0% | 1,339 | 5,859 | +338% | 0 | 0 | — |
case-20 | fail→pass | 15,267 | 14,687 | -4% | 1 | 1 | 0% | 2,837 | 8,016 | +183% | 0 | 0 | — |
case-21 | fail→fail | 29,562 | 7,503 | -75% | 1 | 1 | 0% | 6,183 | 5,495 | -11% | 0 | 0 | — |
case-22 | fail→fail | 27,580 | 54,806 | +99% | 1 | 1 | 0% | 6,172 | 6,497 | +5% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 10 counted toward the lift figure. The other 12 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +36 percentage points is the difference between those two pass rates over the 10 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.