ThomasMoreAI/legal-skills-open / br/litigation/skills/juriscan-brunobracaioli
juriscan-brunobracaioli
VerifiedAnálise forense de processos judiciais brasileiros. Extrai, classifica e cruza peças processuais, detecta contradições, calcula prazos CPC e gera relatórios de risco com exportação para Obsidian. Use quando mencionar: processo judicial, petição, sentença, acórdão, contradições, prazos, timeline, análise forense, Obsidian.
scripts/legacy/risk_scorer.py
View on GitHub#!/usr/bin/env python3
"""
risk_scorer.py — Litigation risk scoring.
DEPRECATED (Phase 5 Step 5.1, 2026-04-11): moved to scripts/legacy/. Still
consumed by SKILL.md when `--pipeline=legacy` is explicit. The agents
pipeline uses juriscan-advogado-autor + juriscan-advogado-reu (two-polo
perspectives) instead. Removal scheduled 3 releases after Phase 6.4 flip.
Computes structural risk indicators from the analysis JSON.
Strategic analysis (Claude-dependent) is handled via prompt templates.
Usage:
python3 risk_scorer.py --analysis analyzed.json --output risk.json
"""
from __future__ import annotations
import argparse
import json
import os
import re
import sys
import warnings as _warnings
_warnings.warn(
"scripts.legacy.risk_scorer is deprecated; prefer perspectives.autor / "
"perspectives.reu from the agents pipeline.",
DeprecationWarning,
stacklevel=2,
)
sys.path.insert(0, os.path.dirname(__file__))
# risk_scorer previously imported from scripts/utils — after the move to
# scripts/legacy/, point to the parent scripts/ so utils.monetary still works.
sys.path.insert(0, os.path.join(os.path.dirname(__file__), ".."))
from utils.monetary import normalize_brl
def score_procedural_risk(analysis: dict) -> dict:
"""Score procedural risk based on structural indicators.
Checks:
- Missing expected pieces
- Prazo compliance (from prazos_calculados if available)
- Integrity anomalies
- Chronological consistency
"""
score = 10.0 # Start perfect, deduct
factors = []
chunks = analysis.get('chunks', [])
labels = [c.get('label', '') for c in chunks]
# Missing contestação when petição + sentença exist
if 'PETIÇÃO INICIAL' in labels and 'SENTENÇA' in labels and 'CONTESTAÇÃO' not in labels:
score -= 2.0
factors.append({
'fator': 'Contestação ausente — possível revelia',
'impacto': -2.0,
'peca_fonte': 'Análise de peças',
})
# Prazo violations
prazos = analysis.get('prazos_calculados', [])
vencidos = [p for p in prazos if p.get('status') == 'vencido']
if vencidos:
penalty = min(len(vencidos) * 2.0, 5.0)
score -= penalty
factors.append({
'fator': f'{len(vencidos)} prazo(s) vencido(s)',
'impacto': -penalty,
'peca_fonte': 'Cálculo de prazos',
})
# Integrity issues
integrity = analysis.get('integrity_report', {})
anomalies = integrity.get('anomalies', [])
high_impact = [a for a in anomalies if a.get('impacto') == 'ALTO']
if high_impact:
penalty = min(len(high_impact) * 1.5, 3.0)
score -= penalty
factors.append({
'fator': f'{len(high_impact)} anomalia(s) de alto impacto na integridade documental',
'impacto': -penalty,
'peca_fonte': 'Relatório de integridade',
})
# Low OCR confidence
ocr_scores = integrity.get('ocr_scores', [])
low_ocr = [s for s in ocr_scores if s.get('score', 1) < 0.5]
if low_ocr:
score -= 1.0
factors.append({
'fator': f'{len(low_ocr)} chunk(s) com OCR confidence < 50%',
'impacto': -1.0,
'peca_fonte': 'Relatório de integridade',
})
return {
'score': round(max(score, 0.0), 1),
'factors': factors,
}
def score_merit_indicators(analysis: dict) -> dict:
"""Score merit indicators from structural data.
Analyzes: contradiction count/severity, evidence presence, jurisprudence alignment.
"""
score = 5.0 # Neutral start
favorable = []
unfavorable = []
chunks = analysis.get('chunks', [])
# Contradictions analysis
contradictions = analysis.get('contradictions', [])
high_contradictions = [c for c in contradictions if c.get('impacto') == 'ALTO']
if high_contradictions:
score -= 1.5
unfavorable.append({
'fator': f'{len(high_contradictions)} contradição(ões) de alto impacto',
'fundamentacao': '; '.join(c.get('descricao', '')[:100] for c in high_contradictions[:3]),
})
# Evidence: laudo pericial present?
has_laudo = any(c.get('label') == 'LAUDO PERICIAL' for c in chunks)
if has_laudo:
score += 1.0
favorable.append({
'fator': 'Laudo pericial presente no processo',
'fundamentacao': 'Prova técnica disponível para fundamentação',
})
# Evidence: ata de audiência (witnesses)
has_audiencia = any(c.get('label') == 'ATA DE AUDIÊNCIA' for c in chunks)
if has_audiencia:
score += 0.5
favorable.append({
'fator': 'Audiência de instrução realizada',
'fundamentacao': 'Prova oral produzida',
})
# Jurisprudence: binding precedents
all_binding = []
for c in chunks:
all_binding.extend(c.get('binding_precedents', []))
favoraveis = [b for b in all_binding if b.get('alinhamento') == 'FAVORAVEL_AUTOR']
desfavoraveis = [b for b in all_binding if b.get('alinhamento') == 'FAVORAVEL_REU']
if favoraveis:
score += min(len(favoraveis) * 0.5, 2.0)
favorable.append({
'fator': f'{len(favoraveis)} precedente(s) vinculante(s) favorável(is)',
'fundamentacao': ', '.join(b.get('numero', '?') for b in favoraveis[:3]),
})
if desfavoraveis:
score -= min(len(desfavoraveis) * 0.5, 2.0)
unfavorable.append({
'fator': f'{len(desfavoraveis)} precedente(s) vinculante(s) desfavorável(is)',
'fundamentacao': ', '.join(b.get('numero', '?') for b in desfavoraveis[:3]),
})
# Sentença result (if available)
for c in chunks:
if c.get('label') == 'SENTENÇA' and c.get('decisao'):
decisao = c['decisao'].lower()
if 'procedente' in decisao and 'improcedente' not in decisao:
score += 1.5
favorable.append({
'fator': 'Sentença procedente (total ou parcial)',
'fundamentacao': c['decisao'][:150],
'peca_fonte': 'SENTENÇA',
})
return {
'score': round(max(min(score, 10.0), 0.0), 1),
'favorable_factors': favorable,
'unfavorable_factors': unfavorable,
}
def score_monetary_exposure(analysis: dict) -> dict:
"""Calculate monetary exposure from available data."""
chunks = analysis.get('chunks', [])
max_exposure = 0.0
likely_value = 0.0
min_value = 0.0
for c in chunks:
valores = c.get('valores', {})
# Maximum: valor da causa or sum of pedidos
causa = normalize_brl(valores.get('causa', '') or '')
if causa and causa > max_exposure:
max_exposure = causa
# Likely: condenação value (from sentença/acórdão)
cond = normalize_brl(valores.get('condenacao', '') or '')
if cond and c.get('label') in ('SENTENÇA', 'ACÓRDÃO'):
likely_value = cond
# Values from "outros"
for outro in valores.get('outros', []):
val = normalize_brl(outro.get('valor', '') or '')
if val:
if val > min_value:
min_value = val
# If no condenação, estimate likely as 60% of max
if not likely_value and max_exposure:
likely_value = max_exposure * 0.6
# Min is the smallest acknowledged value or 30% of likely
if not min_value:
min_value = likely_value * 0.3 if likely_value else 0.0
# Costs estimation
honorarios = likely_value * 0.15 if likely_value else max_exposure * 0.10
custas = max_exposure * 0.01 if max_exposure else 0.0
# Ensure invariant: likely_range.min <= likely_range.max.
# Previously min_value could exceed likely_value when "outros" valores
# (e.g. aggregated pedidos) were larger than the actual condenação,
# producing a nonsense range with min > max.
range_min, range_max = sorted([min_value, likely_value])
assert range_min <= range_max, (
f"likely_range invariant violated: min={range_min}, max={range_max}"
)
def fmt(v: float) -> str | None:
if v <= 0:
return None
return f"R$ {v:,.2f}".replace(',', 'X').replace('.', ',').replace('X', '.')
return {
'max_exposure': fmt(max_exposure),
'likely_range': {
'min': fmt(range_min),
'max': fmt(range_max),
},
'costs': {
'custas_estimadas': fmt(custas),
'honorarios_sucumbencia': fmt(honorarios),
},
}
def generate_risk_report(analysis: dict) -> dict:
"""Generate composite risk report."""
procedural = score_procedural_risk(analysis)
merit = score_merit_indicators(analysis)
monetary = score_monetary_exposure(analysis)
# Weighted composite score (from author's perspective)
overall = procedural['score'] * 0.3 + merit['score'] * 0.5 + (procedural['score'] * 0.2)
overall = round(overall, 1)
if overall < 4.0:
level = 'ALTO'
elif overall < 7.0:
level = 'MÉDIO'
else:
level = 'BAIXO'
return {
'risk_level': level,
'overall_score': overall,
'procedural_risk': procedural,
'merit_indicators': merit,
'monetary_exposure': monetary,
'strategic_recommendations': [], # Populated by Claude via prompt
}
def main():
parser = argparse.ArgumentParser(description='Litigation risk scorer')
parser.add_argument('--analysis', '-a', required=True, help='Path to analyzed.json')
parser.add_argument('--output', '-o', help='Output path')
args = parser.parse_args()
with open(args.analysis, 'r', encoding='utf-8') as f:
analysis = json.load(f)
report = generate_risk_report(analysis)
output_path = args.output or os.path.join(os.path.dirname(args.analysis), 'risk.json')
with open(output_path, 'w', encoding='utf-8') as f:
json.dump(report, f, ensure_ascii=False, indent=2)
print(f"\nRisk Assessment:")
print(f" Level: {report['risk_level']}")
print(f" Score: {report['overall_score']}/10")
print(f" Procedural: {report['procedural_risk']['score']}/10")
print(f" Merit: {report['merit_indicators']['score']}/10")
print(f" Output: {output_path}")
if __name__ == '__main__':
main()
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |