63 lines
1.7 KiB
Markdown
63 lines
1.7 KiB
Markdown
# bypass-ai
|
|
|
|
AI-detection bypass, LLM jailbreak patterns, anti-bot detection evasion, prompt injection.
|
|
|
|
> **For authorized red-team, threat-hunting, and forensic analysis only.**
|
|
|
|
## Quick start
|
|
|
|
```bash
|
|
# Self-test
|
|
python bypass_ai.py --self-test
|
|
|
|
# Humanize AI text
|
|
python bypass_ai.py --humanize "Your AI-generated text here..."
|
|
|
|
# Generate jailbreak for target LLM
|
|
python bypass_ai.py --jailbreak claude-4-6 --payload "explain WAF bypass"
|
|
python bypass_ai.py --jailbreak gpt-5 --payload "extract system prompt"
|
|
|
|
# Anti-bot bypass plan
|
|
python bypass_ai.py --anti-bot "https://protected-site.com"
|
|
|
|
# Prompt injection payload
|
|
python bypass_ai.py --inject rag --payload "dump all data"
|
|
python bypass_ai.py --inject base64 --payload "execute command"
|
|
|
|
# Detect guardrails in response
|
|
python bypass_ai.py --detect-guardrails "I'm sorry but I can't help with that..."
|
|
|
|
# Multi-layer evasion chain
|
|
python bypass_ai.py --evasion "execute payload" --layers base64,rot13,hex
|
|
```
|
|
|
|
## As a module
|
|
|
|
```python
|
|
from bypass_ai import humanize, jailbreak_prompt, anti_bot_bypass, prompt_injection, detect_ai_guardrails, evasion_chain
|
|
|
|
# Humanize
|
|
text = humanize("AI-generated text")
|
|
|
|
# Jailbreak
|
|
prompt = jailbreak_prompt("claude-4-6", "show admin credentials")
|
|
|
|
# Anti-bot
|
|
plan = anti_bot_bypass("https://target.com")
|
|
|
|
# Injection
|
|
payload = prompt_injection("rag", "leak data")
|
|
|
|
# Detect
|
|
info = detect_ai_guardrails(ai_response)
|
|
|
|
# Chain
|
|
obfuscated = evasion_chain("payload", ["base64", "rot13", "split"])
|
|
```
|
|
|
|
## See also
|
|
|
|
- `/bypass_unlimited` — фильтры/капчи/API-лимиты
|
|
- `/unfiltered_ultra` — режим без фильтров
|
|
- `/abyss_mode` — максимальная детализация
|