lovarch-cli 0.2.1__py3-none-any.whl

lovarch_cli/squad/scripts/validate-squad.py

@@ -0,0 +1,383 @@
+"""
+validate-squad.py · AIOS Quality Gate Validator
+
+Runs squad-checklist v3.1 (Tier 1-4) against squad architettura-progetto
+and outputs a quality score (target: 10/10).
+
+Usage:
+    python3 validate-squad.py [--verbose]
+
+Exit codes:
+    0 = PASS (score ≥7.0)
+    1 = FAIL (score <7.0)
+    2 = ERROR (validation crashed)
+"""
+import json
+import re
+import sys
+from pathlib import Path
+from typing import Dict, List, Tuple
+
+
+SQUAD_DIR = Path(__file__).parent.parent
+VERBOSE = "--verbose" in sys.argv
+
+
+# ============================================================================
+# TIER 1 · STRUCTURE (BLOCKING)
+# ============================================================================
+
+def check_tier_1() -> Tuple[float, List[Dict]]:
+    """Returns (score 0-10, list of check results)"""
+    checks = []
+
+    # 1.1 config.yaml exists + parseable
+    config_path = SQUAD_DIR / "config.yaml"
+    config_exists = config_path.exists()
+    checks.append({"id": "1.1", "name": "config.yaml exists", "pass": config_exists, "severity": "BLOCKING"})
+
+    config_text = config_path.read_text() if config_exists else ""
+
+    # 1.1 Required fields
+    required_fields = ["name:", "version:", "description:", "entry_agent:"]
+    for field in required_fields:
+        present = field in config_text
+        checks.append({
+            "id": f"1.1.{field}",
+            "name": f"config field {field}",
+            "pass": present,
+            "severity": "BLOCKING",
+        })
+
+    # 1.1 author
+    author_present = "author:" in config_text
+    checks.append({"id": "1.1.author", "name": "config has author", "pass": author_present, "severity": "BLOCKING"})
+
+    # 1.1 slashPrefix
+    slash_present = "slashPrefix:" in config_text
+    checks.append({"id": "1.1.slashPrefix", "name": "config has slashPrefix", "pass": slash_present, "severity": "BLOCKING"})
+
+    # 1.4 Directory structure
+    dirs = ["agents", "tasks", "checklists", "data", "templates", "workflows", "scripts"]
+    for d in dirs:
+        d_path = SQUAD_DIR / d
+        exists = d_path.exists() and d_path.is_dir()
+        checks.append({"id": f"1.4.{d}", "name": f"dir {d}/", "pass": exists, "severity": "BLOCKING"})
+
+    # 1.4 Min agents count (>=1)
+    agents_dir = SQUAD_DIR / "agents"
+    agents_count = len(list(agents_dir.glob("*.md"))) if agents_dir.exists() else 0
+    checks.append({
+        "id": "1.4.agents_count",
+        "name": f"agents count >=1 (actual: {agents_count})",
+        "pass": agents_count >= 1,
+        "severity": "BLOCKING",
+    })
+
+    # 1.4 Min tasks count (>=1 per AIOS standard)
+    tasks_dir = SQUAD_DIR / "tasks"
+    tasks_count = len(list(tasks_dir.glob("*.md"))) if tasks_dir.exists() else 0
+    checks.append({
+        "id": "1.4.tasks_count",
+        "name": f"tasks count >=1 (actual: {tasks_count})",
+        "pass": tasks_count >= 1,
+        "severity": "BLOCKING",
+    })
+
+    # 1.5 Cross-references
+    central_doc = SQUAD_DIR / "data" / "architettura-progetto-rules.md"
+    central_exists = central_doc.exists()
+    checks.append({"id": "1.5.central_doc", "name": "central document exists", "pass": central_exists, "severity": "BLOCKING"})
+
+    changelog = SQUAD_DIR / "data" / "CHANGELOG.md"
+    changelog_exists = changelog.exists()
+    checks.append({"id": "1.5.changelog", "name": "CHANGELOG.md exists", "pass": changelog_exists, "severity": "BLOCKING"})
+
+    handoff_card = SQUAD_DIR / "data" / "handoff-card-template.md"
+    handoff_card_exists = handoff_card.exists()
+    checks.append({"id": "1.5.handoff_card", "name": "handoff-card-template exists", "pass": handoff_card_exists, "severity": "BLOCKING"})
+
+    handoff_qa = SQUAD_DIR / "checklists" / "handoff-quality-gate.md"
+    handoff_qa_exists = handoff_qa.exists()
+    checks.append({"id": "1.5.handoff_qa", "name": "handoff-quality-gate exists", "pass": handoff_qa_exists, "severity": "BLOCKING"})
+
+    # 1.6 Security scan (no hardcoded secrets in YAML)
+    secrets_pattern = re.compile(r"(api[_-]?key|secret|password)\s*[:=]\s*['\"][^'\"]{20,}", re.IGNORECASE)
+    has_secrets = bool(secrets_pattern.search(config_text))
+    checks.append({"id": "1.6.no_secrets", "name": "no hardcoded secrets", "pass": not has_secrets, "severity": "BLOCKING"})
+
+    # Score
+    total = len(checks)
+    passed = sum(1 for c in checks if c["pass"])
+    score = (passed / total) * 10
+    return score, checks
+
+
+# ============================================================================
+# TIER 2 · COVERAGE (BLOCKING)
+# ============================================================================
+
+def check_tier_2() -> Tuple[float, List[Dict]]:
+    checks = []
+
+    # Tier 0 orchestrator exists
+    chief_exists = (SQUAD_DIR / "agents" / "progetto-chief.md").exists()
+    checks.append({"id": "2.1.chief", "name": "Tier 0 orchestrator exists", "pass": chief_exists, "severity": "BLOCKING"})
+
+    # Auditor input
+    auditor_exists = (SQUAD_DIR / "agents" / "auditor-input.md").exists()
+    checks.append({"id": "2.2.auditor", "name": "auditor-input exists", "pass": auditor_exists, "severity": "BLOCKING"})
+
+    # Tier 1 specialists ≥5
+    tier1_agents = [
+        "briefing-architect", "regolatorio-it", "concept-designer", "cad-engineer",
+        "bim-engineer", "computo-engineer", "capitolato-writer", "pratiche-it",
+        "contratto-architect", "energy-prelim", "deliverable-builder",
+    ]
+    tier1_present = sum(1 for a in tier1_agents if (SQUAD_DIR / "agents" / f"{a}.md").exists())
+    checks.append({
+        "id": "2.3.tier1_count",
+        "name": f"Tier 1 specialists ≥5 (actual: {tier1_present})",
+        "pass": tier1_present >= 5,
+        "severity": "BLOCKING",
+    })
+
+    # Tier 2 QA agents
+    qa_agents = ["quality-misure", "quality-normativa", "quality-dati", "quality-output"]
+    qa_present = sum(1 for a in qa_agents if (SQUAD_DIR / "agents" / f"{a}.md").exists())
+    checks.append({
+        "id": "2.4.qa_count",
+        "name": f"Tier 2 QA agents ≥4 (actual: {qa_present})",
+        "pass": qa_present >= 4,
+        "severity": "BLOCKING",
+    })
+
+    # Tasks ≥3 (atomic)
+    tasks = list((SQUAD_DIR / "tasks").glob("*.md")) if (SQUAD_DIR / "tasks").exists() else []
+    checks.append({
+        "id": "2.5.tasks_count",
+        "name": f"Tasks ≥3 (actual: {len(tasks)})",
+        "pass": len(tasks) >= 3,
+        "severity": "BLOCKING",
+    })
+
+    # Workflows ≥1
+    workflows = list((SQUAD_DIR / "workflows").glob("*.yaml")) if (SQUAD_DIR / "workflows").exists() else []
+    checks.append({
+        "id": "2.6.workflows_count",
+        "name": f"Workflows ≥1 (actual: {len(workflows)})",
+        "pass": len(workflows) >= 1,
+        "severity": "BLOCKING",
+    })
+
+    # Templates ≥3
+    templates = list((SQUAD_DIR / "templates").glob("*.md")) if (SQUAD_DIR / "templates").exists() else []
+    checks.append({
+        "id": "2.7.templates_count",
+        "name": f"Templates ≥3 (actual: {len(templates)})",
+        "pass": len(templates) >= 3,
+        "severity": "BLOCKING",
+    })
+
+    # Checklists ≥4 (1 handoff + 4 QA)
+    checklists = list((SQUAD_DIR / "checklists").glob("*.md")) if (SQUAD_DIR / "checklists").exists() else []
+    checks.append({
+        "id": "2.8.checklists_count",
+        "name": f"Checklists ≥5 (actual: {len(checklists)})",
+        "pass": len(checklists) >= 5,
+        "severity": "BLOCKING",
+    })
+
+    total = len(checks)
+    passed = sum(1 for c in checks if c["pass"])
+    score = (passed / total) * 10
+    return score, checks
+
+
+# ============================================================================
+# TIER 3 · QUALITY (THRESHOLD 7.0)
+# ============================================================================
+
+REQUIRED_AGENT_SECTIONS = [
+    "ACTIVATION-NOTICE",
+    "IDE-FILE-RESOLUTION",
+    "REQUEST-RESOLUTION",
+    "activation-instructions",
+    "command_loader",
+    "agent:",
+    "persona:",
+    "core_principles",
+    "voice_dna:",
+    "thinking_dna:",
+    "handoff_to:",
+    "output_examples:",
+    "anti_patterns:",
+    "completion_criteria:",
+    "smoke_tests:",
+    "integration:",
+    "greeting:",
+]
+
+
+def check_agent_quality(agent_path: Path) -> Tuple[float, List[Dict]]:
+    """Per-agent AIOS 6-level structure check."""
+    if not agent_path.exists():
+        return 0, [{"id": "missing", "name": f"{agent_path.name} missing", "pass": False, "severity": "BLOCKING"}]
+
+    content = agent_path.read_text()
+    checks = []
+
+    for section in REQUIRED_AGENT_SECTIONS:
+        present = section in content
+        checks.append({
+            "id": f"agent.{section}",
+            "name": f"{agent_path.name} has {section}",
+            "pass": present,
+            "severity": "QUALITY",
+        })
+
+    # Additional quality checks
+    has_smoke_3 = content.count("test_") >= 3
+    checks.append({"id": "agent.smoke_tests_3", "name": f"{agent_path.name} 3 smoke tests", "pass": has_smoke_3, "severity": "QUALITY"})
+
+    has_examples_3 = len(re.findall(r"- input:", content)) >= 3
+    checks.append({"id": "agent.examples_3", "name": f"{agent_path.name} 3 output examples", "pass": has_examples_3, "severity": "QUALITY"})
+
+    has_source_traces = "[SOURCE:" in content or "[Deming" in content or "[Juran" in content or "[English" in content or "[Dodds" in content or "[Schumacher" in content or "[Baldwin" in content or "[Mazria" in content or "signature]" in content
+    checks.append({"id": "agent.source_traces", "name": f"{agent_path.name} has [SOURCE:] traces", "pass": has_source_traces, "severity": "QUALITY"})
+
+    total = len(checks)
+    passed = sum(1 for c in checks if c["pass"])
+    score = (passed / total) * 10
+    return score, checks
+
+
+def check_tier_3() -> Tuple[float, List[Dict]]:
+    agents = list((SQUAD_DIR / "agents").glob("*.md"))
+    all_checks = []
+    scores = []
+
+    for agent_path in agents:
+        score, checks = check_agent_quality(agent_path)
+        scores.append(score)
+        all_checks.extend(checks)
+
+    avg_score = sum(scores) / len(scores) if scores else 0
+    return avg_score, all_checks
+
+
+# ============================================================================
+# TIER 4 · CONTEXTUAL (squad type)
+# ============================================================================
+
+def check_tier_4() -> Tuple[float, List[Dict]]:
+    """Detect squad type · architettura-progetto = Hybrid (process + heuristics)"""
+    checks = []
+    config_text = (SQUAD_DIR / "config.yaml").read_text() if (SQUAD_DIR / "config.yaml").exists() else ""
+
+    has_hub_spoke = "hub_and_spoke" in config_text
+    checks.append({"id": "4.1.hub_spoke", "name": "Hub-and-spoke topology", "pass": has_hub_spoke, "severity": "QUALITY"})
+
+    has_executor_types = "executor_types:" in config_text
+    checks.append({"id": "4.2.executor_types", "name": "Executor types defined", "pass": has_executor_types, "severity": "QUALITY"})
+
+    has_pattern_lib = "pattern_library:" in config_text
+    checks.append({"id": "4.3.pattern_library", "name": "Pattern library defined", "pass": has_pattern_lib, "severity": "QUALITY"})
+
+    has_quality_gates = "quality_gates:" in config_text
+    checks.append({"id": "4.4.quality_gates", "name": "Quality gates referenced", "pass": has_quality_gates, "severity": "QUALITY"})
+
+    # Mind clones traceable
+    mind_clones = ["quality-misure", "quality-normativa", "quality-dati", "quality-output", "concept-designer", "bim-engineer", "energy-prelim"]
+    clones_with_source = 0
+    for c in mind_clones:
+        cpath = SQUAD_DIR / "agents" / f"{c}.md"
+        if cpath.exists() and "based_on:" in cpath.read_text():
+            clones_with_source += 1
+    checks.append({
+        "id": "4.5.mind_clones",
+        "name": f"Mind clones documented ({clones_with_source}/7)",
+        "pass": clones_with_source >= 7,
+        "severity": "QUALITY",
+    })
+
+    total = len(checks)
+    passed = sum(1 for c in checks if c["pass"])
+    score = (passed / total) * 10
+    return score, checks
+
+
+# ============================================================================
+# RUNNER
+# ============================================================================
+
+def main():
+    print("\n" + "=" * 70)
+    print("  Squad Validation · architettura-progetto")
+    print("  Applying squad-checklist v3.1 · AIOS Standard")
+    print("=" * 70 + "\n")
+
+    t1_score, t1_checks = check_tier_1()
+    t2_score, t2_checks = check_tier_2()
+    t3_score, t3_checks = check_tier_3()
+    t4_score, t4_checks = check_tier_4()
+
+    # Final score (weighted: T3 × 0.5 + T1 × 0.2 + T2 × 0.2 + T4 × 0.1)
+    final = (t3_score * 0.5) + (t1_score * 0.2) + (t2_score * 0.2) + (t4_score * 0.1)
+
+    print(f"Tier 1 · Structure:    {t1_score:5.2f}/10  ({sum(1 for c in t1_checks if c['pass'])}/{len(t1_checks)} checks)")
+    print(f"Tier 2 · Coverage:     {t2_score:5.2f}/10  ({sum(1 for c in t2_checks if c['pass'])}/{len(t2_checks)} checks)")
+    print(f"Tier 3 · Quality:      {t3_score:5.2f}/10  (per-agent average · {len(list((SQUAD_DIR/'agents').glob('*.md')))} agents)")
+    print(f"Tier 4 · Contextual:   {t4_score:5.2f}/10  ({sum(1 for c in t4_checks if c['pass'])}/{len(t4_checks)} checks)")
+    print(f"")
+    print(f"  ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━")
+    print(f"  FINAL SCORE:  {final:5.2f}/10")
+    print(f"  ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━")
+    print(f"")
+
+    if final >= 9.0:
+        print("  Verdict: EXCELLENCE 🏆")
+    elif final >= 7.0:
+        print("  Verdict: PASS ✓")
+    else:
+        print("  Verdict: FAIL ✗")
+    print("")
+
+    if VERBOSE:
+        print("\n--- Tier 1 · Structure (BLOCKING) ---")
+        for c in t1_checks:
+            mark = "✓" if c["pass"] else "✗"
+            print(f"  {mark} [{c['id']}] {c['name']}")
+
+        print("\n--- Tier 2 · Coverage (BLOCKING) ---")
+        for c in t2_checks:
+            mark = "✓" if c["pass"] else "✗"
+            print(f"  {mark} [{c['id']}] {c['name']}")
+
+        # Tier 3 · group by agent
+        print("\n--- Tier 3 · Quality (per agent) ---")
+        agents = list((SQUAD_DIR / "agents").glob("*.md"))
+        for agent_path in agents:
+            score, checks = check_agent_quality(agent_path)
+            print(f"  {agent_path.stem}: {score:.1f}/10  ({sum(1 for c in checks if c['pass'])}/{len(checks)})")
+
+        print("\n--- Tier 4 · Contextual ---")
+        for c in t4_checks:
+            mark = "✓" if c["pass"] else "✗"
+            print(f"  {mark} [{c['id']}] {c['name']}")
+
+        # Failed only
+        all_failed = [c for c in (t1_checks + t2_checks + t4_checks) if not c["pass"]]
+        if all_failed:
+            print(f"\n--- {len(all_failed)} FAILED CHECKS ---")
+            for c in all_failed:
+                print(f"  ✗ [{c['id']}] {c['name']}")
+
+    if final >= 7.0:
+        sys.exit(0)
+    else:
+        sys.exit(1)
+
+
+if __name__ == "__main__":
+    main()

lovarch_cli/squad/tasks/audit-input.md

@@ -0,0 +1,146 @@
+# Task: audit-input
+
+> **Pattern:** AP-TP-001 (Atomic Task Anatomy · 8 mandatory fields)
+> **Executor:** @auditor-input (functional · Tier 0)
+> **Squad:** architettura-progetto
+
+---
+
+## task_name
+Audit input completeness · 18-item checklist gate
+
+## status
+ACTIVE · stable · v1.0
+
+## responsible_executor
+- **agent**: @auditor-input
+- **executor_type**: AP-EP-002 (Agent Executor)
+- **fallback**: @progetto-chief escalation if agent timeout
+
+## execution_type
+**Synchronous · Pre-flight gate** · Blocks workflow if FAIL
+
+## input
+```yaml
+required:
+  brief_path: "Path to briefing-cliente.md"
+  dwg_path: "Path to stato-attuale.dxf"
+  photos_dir: "Directory with stato attuale photos"
+  cliente_data:
+    nome: string
+    cognome: string
+    codice_fiscale: string  # 16-char Italian CF
+    email: string
+    telefono: string
+  studio_data:
+    architetto: string
+    n_ordine: string
+    piva: string
+    pec: string
+  valore_opera: number  # EUR
+optional:
+  visura_path: "Catastal record PDF"
+```
+
+## output
+```yaml
+file: "input_validation.json"
+schema:
+  validation_id: uuid
+  status: "PASS | FAIL"
+  missing: array[string]  # item ids that failed
+  warnings: array[string]
+  extracted_data:
+    client_name: string
+    client_cf_primary: string
+    address: string
+    geocoded:
+      lat: number
+      lon: number
+      comune: string
+      postcode: string
+    project_value: number
+    studio: object
+```
+
+## action_items
+1. **A1** Read briefing-cliente.md · verify char count ≥500
+2. **A2** Verify briefing has ≥3 of 12 sezioni UNI 11337
+3. **A3** Extract budget numerico (regex /€\s?[\d,.]+/)
+4. **A4** Extract timeline indicators
+5. **A5** Verify briefing has cliente expectations section
+6. **B1** Run `ezdxf.readfile(dwg)` · entities count > 0
+7. **B2** Iterate photos · verify ≥3 JPG/PNG · resolution ≥800x600
+8. **B3** Check visura PDF if present
+9. **C1** Verify cliente nome+cognome non-empty
+10. **C2-C3** Verify CF format /^[A-Z0-9]{16}$/ · Italian checksum algorithm
+11. **C4** Verify email regex
+12. **C5** Mapbox API call · geocode address · features.length > 0
+13. **D1** Verify architetto nome
+14. **D2** Verify n. Ordine (digit string)
+15. **D3** Verify P.IVA studio (11-digit + checksum)
+16. **D4** Verify PEC format
+17. **E1** Verify valore_opera numeric > 0
+18. **Final** Build extracted_data + verdict
+
+## acceptance_criteria
+- [ ] All 18 items checked (no skipping)
+- [ ] If 7/7 critici (A1, B1, B2, C2-C3, C5, D3, E1) PASS → status=PASS
+- [ ] If ANY critico FAIL → status=FAIL · workflow halts
+- [ ] JSON output valid against schema
+- [ ] If PASS · extracted_data populated for downstream
+- [ ] If FAIL · missing[] lists exact item_ids
+- [ ] Verdict announcement returned to @progetto-chief
+
+## dependencies
+- **APIs:**
+  - Mapbox Geocoding API (item C5 critical)
+  - Italian CF checksum library (item C2-C3)
+- **Files:**
+  - All input files (brief, dwg, photos, etc.)
+- **Agents:**
+  - @progetto-chief (invoker · returns to)
+
+## templates
+- `~/projects/{slug}/01-input/` (expected input directory structure)
+- input_validation.json schema (above)
+
+## quality_gate
+- **Gate:** QG-AP-1.1 (Input Validation Gate)
+- **Threshold:** 7/7 CRITICI must PASS (A1, B1, B2, C2-C3, C5, D3, E1)
+- **Reviewer:** @progetto-chief on receipt
+
+## handoff
+- **From:** @progetto-chief (outbound card)
+- **To:** @progetto-chief (inbound card with validation_id + verdict)
+- **Required announcement:** "Ritorno al @progetto-chief. Audit completato — verdict: {PASS|FAIL}."
+
+## veto_conditions
+- Briefing < 500 chars → FAIL (A1)
+- DWG corrupted (ezdxf raises) → FAIL (B1)
+- Photos < 3 → FAIL (B2)
+- Address not geocodable → FAIL (C5)
+- Invalid CF → FAIL (C2-C3)
+- Invalid P.IVA → FAIL (D3)
+- Valore opera ≤0 → FAIL (E1)
+
+## estimated_time
+**60-90 seconds** (Mapbox API call dominant)
+
+## output_example
+```json
+{
+  "validation_id": "v_a7f4b2",
+  "status": "PASS",
+  "missing": [],
+  "warnings": [],
+  "extracted_data": {
+    "client_name": "Marco Rossini & Giulia Bianchi",
+    "client_cf_primary": "RSSMRC83A15F205X",
+    "address": "Via Fiori Chiari 17, 20121 Milano",
+    "geocoded": {"lat": 45.471823, "lon": 9.184828, "comune": "Milano", "postcode": "20121"},
+    "project_value": 180000,
+    "studio": {"nome": "Pablo Ruan", "ordine_n": "XXXX", "piva": "01234567890"}
+  }
+}
+```

lovarch_cli/squad/tasks/compute-metric.md

@@ -0,0 +1,105 @@
+# Task: compute-metric
+
+> **Pattern:** AP-TP-001
+> **Executor:** @computo-engineer (functional · Tier 1 · critical: dati_zero_tolerance)
+> **Squad:** architettura-progetto
+
+---
+
+## task_name
+Compute metric estimative · Prezzario Lombardia 2025 · IVA 10%
+
+## status
+ACTIVE · stable · v1.0
+
+## responsible_executor
+- **agent**: @computo-engineer
+- **executor_type**: AP-EP-002 (Agent + Python worker)
+- **worker**: xlsxwriter, openpyxl, pdfplumber
+
+## execution_type
+**Synchronous after BIM** · Output feeds @capitolato-writer + @quality-dati
+
+## input
+```yaml
+required:
+  quantitativi_json: "From @bim-engineer"
+  prezzario_path: "data/prezzario-lombardia-sample.json"
+optional:
+  dei_plus_subscription: bool  # fallback per voci mancanti
+  ec3_for_epd: bool  # for EPD tracking
+```
+
+## output
+```yaml
+files:
+  - "05-impresa/computo-metrico.xlsx"
+  - "05-impresa/computo-metrico.pdf"
+  - "05-impresa/quadro-economico.pdf"
+  - "05-impresa/lista-materiali-EPDs.xlsx"
+metrics:
+  voci_count: number
+  totale_lavori_eur: number
+  iva_eur: number
+  totale_iva_inclusa: number
+  cam_compliance_percent: number
+```
+
+## action_items
+1. Read quantitativi.json (volumi muri, aree pavimenti, count, etc.)
+2. Match each quantity to Prezzario Lombardia voce (semantic + code)
+3. For unmatched: query DEI fallback OR mark [VERIFY-CUSTOM]
+4. Compute Q × prezzo_unitario per voce
+5. Aggregate per categoria DEI (demolizioni, murature, impianti, etc.)
+6. Apply IVA 10% (ristrutturazione · DPR 633/72)
+7. Build quadro economico (lavori + onorari + oneri + IVA + imprevisti)
+8. Track CAM 2025 compliance per voce (target ≥80%)
+9. Cross-reference EC3 for EPDs in materials list
+10. Generate xlsx with formulas (SUM, IVA calc)
+11. Generate PDF version
+12. Self-verify: sum quantitativi IFC = sum computo (diff <2%)
+
+## acceptance_criteria
+- [ ] ≥100 voci with codice Prezzario or [VERIFY-CUSTOM]
+- [ ] Aggregazione per categoria DEI completa
+- [ ] IVA 10% applied (NOT 22%)
+- [ ] Quadro economico generato
+- [ ] Cross-check IFC quantitativi · diff <2%
+- [ ] CAM 2025 tracking ≥80% target
+- [ ] xlsx with working formulas
+- [ ] PDF readable
+
+## dependencies
+- **Libraries:**
+  - xlsxwriter (xlsx generation)
+  - openpyxl (template editing)
+  - pdfplumber (parse Prezzario PDF if needed)
+- **APIs:**
+  - DEI PLUS (optional · fallback)
+  - EC3 Building Transparency (EPDs)
+  - Gemini 3.1 Pro (gemini-3.1-pro-preview) (semantic mapping descrizione → voce)
+- **Data:**
+  - Prezzario Regione Lombardia 2025 cached JSON
+  - quantitativi.json (@bim-engineer)
+
+## quality_gate
+- **Gate:** QG-AP-1.4 (Cross-Doc Data Gate)
+- **Reviewer:** @quality-dati
+- **Threshold:** Cross-check IFC = computo · diff <2%
+
+## handoff
+- **From:** @progetto-chief
+- **To:** @progetto-chief → @capitolato-writer (uses computo) + @quality-dati (verifies)
+- **Required announcement:** "Ritorno al @progetto-chief. Computo · {n} voci · totale € {X}."
+
+## veto_conditions
+- Totale computo ≠ sum voci > 0.5% → halt
+- IVA 22% applied → halt (errata · ristrutturazione = 10%)
+- Voci senza prezzo → halt
+- Diff vs IFC quantitativi >2% → halt + flag @quality-dati REJECT predicted
+
+## estimated_time
+**45-60 seconds**
+
+## output_example
+See `@computo-engineer.md` output_examples · 124 voci · €162,327 · CAM 87% · cross-check pass.