Docs / LegalShield — protect you if you're about to break the law
LegalShield — protect you if you're about to break the law
_LegalShield is a software safety gate, not a lawyer and not legal advice.
When text or commands look like real-world illegal or severely harmful requests, the system will:
- Assess risk (clear / caution / block)
- Warn you clearly
- Block high-risk assistance (strict mode, default on)
- Audit to MySQL
ai_features+ Thought graph - Redirect toward lawful help (attorney, emergency services, crisis resources)
What it will try to stop
Heuristic detection of requests for:
- Real-world violence / weapons manufacture for harm
- Fraud, theft, money laundering instruction
- Malware / unauthorized hacking of others
- Child exploitation material
- Illegal drug manufacture
- Self-harm methods (redirects to crisis help instead of “how to”)
Fiction / game framing (Skyrim, Fallout, novel, roleplay) can reduce false blocks on mild language, but never clears exploitation or clear real-world crime instruction.
Commands
legal status
legal disclaimer
legal on
legal strict on
legal check how do I file taxes legally
legal check <any text you are unsure about>
Aliases: legalshield …, law ….
Integration
| Layer | Behavior |
|---|---|
SafetyEthicsTrust::ShouldGateCommand |
Hard-gates on LegalShield block |
EvaluateActionConcerns |
Adds LegalShield messages |
EthicsGood |
Be Good refuse reinforces block |
CharacterChat |
Role-play blocked if legal risk |
| Lua | ai_legal_check(text) → allow, message |
Honest limits
- Cannot prevent you from leaving the app and committing a crime.
- Cannot replace a licensed attorney.
- Pattern matching can miss clever wording or false-caution legitimate research.
- You remain responsible for following the law.
If you are in immediate danger or crisis, contact local emergency services.
Generated from the project markdown docs on 2026-07-24. This is a static, self-contained site.