Uploaded October 2025 | Updated September 2026, 1 day ago
Jailbreaking LLMs is fun
A backend process restricts the model from writing certain parts of the system prompt inside file. To extract more data, you need to ask for it inside a code block, making sure to tell the model to replace characters to prevent execution of example tool calls defined in its system prompt.
github.com/0xeb/TheBigPromptLibrary/pull/40
Jailbreaking LLMs is fun
A backend process restricts the model from writing certain parts of the system prompt inside file. To extract more data, you need to ask for it inside a code block, making sure to tell the model to replace characters to prevent execution of example tool calls defined in its system prompt.
github.com/0xeb/TheBigPromptLibrary/pull/40
![[OLD] Test dune stab TV developpée par moi en 2008-2009](https://i.ytimg.com/vi/bHgirXxmLhY/mqdefault.jpg "[OLD] Test dune stab TV developpée par moi en 2008-2009
[OLD] Test dune stab TV developpée par moi en 2008-2009")
