8de71d6671
Three-layer defense for AI narrator output: Layer 1: Game-context system prompts with per-game vocabulary whitelists - Morrowind/Skyrim profiles treat mature themes as game mechanics - Whitelisted terms (Skooma, slave, etc.) replaced before guard check Layer 2: Real-time output filter via Llama Guard (Ollama) - llama-guard3:1b for <30ms latency per sentence - Regex fallback when guard model unavailable (graceful degradation) - On fail → contextual fallback narration per scene type Layer 3: Per-game moderation profiles with threshold tuning - Configurable confidence thresholds per game - Low-confidence flags pass through (prevents over-filtering) New files: - src/infrastructure/guards/ — moderation pipeline module - config/moderation.yaml — per-game profile configuration - tests/infrastructure/test_moderation.py — 32 unit tests Fixes #1056 Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
57 lines
1.8 KiB
Python
57 lines
1.8 KiB
Python
"""Load game moderation profiles from config/moderation.yaml.
|
|
|
|
Falls back to hardcoded defaults if the YAML file is missing or malformed.
|
|
"""
|
|
|
|
import logging
|
|
from pathlib import Path
|
|
|
|
from infrastructure.guards.moderation import GameProfile
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
|
|
def load_profiles(config_path: Path | None = None) -> dict[str, GameProfile]:
|
|
"""Load game moderation profiles from YAML config.
|
|
|
|
Args:
|
|
config_path: Path to moderation.yaml. Defaults to config/moderation.yaml.
|
|
|
|
Returns:
|
|
Dict mapping game_id to GameProfile.
|
|
"""
|
|
path = config_path or Path("config/moderation.yaml")
|
|
|
|
if not path.exists():
|
|
logger.info("Moderation config not found at %s — using defaults", path)
|
|
return {}
|
|
|
|
try:
|
|
import yaml
|
|
except ImportError:
|
|
logger.warning("PyYAML not installed — using default moderation profiles")
|
|
return {}
|
|
|
|
try:
|
|
data = yaml.safe_load(path.read_text())
|
|
except Exception as exc:
|
|
logger.error("Failed to parse moderation config: %s", exc)
|
|
return {}
|
|
|
|
profiles: dict[str, GameProfile] = {}
|
|
for game_id, profile_data in data.get("profiles", {}).items():
|
|
try:
|
|
profiles[game_id] = GameProfile(
|
|
game_id=game_id,
|
|
display_name=profile_data.get("display_name", game_id),
|
|
vocabulary_whitelist=profile_data.get("vocabulary_whitelist", []),
|
|
context_prompt=profile_data.get("context_prompt", ""),
|
|
threshold=float(profile_data.get("threshold", 0.8)),
|
|
fallbacks=profile_data.get("fallbacks", {}),
|
|
)
|
|
except Exception as exc:
|
|
logger.warning("Invalid profile '%s': %s", game_id, exc)
|
|
|
|
logger.info("Loaded %d moderation profiles from %s", len(profiles), path)
|
|
return profiles
|