This idea didn't go over well and was adopted by larger devs that have social media bases. It probably won't be updated as it's not worth my time and resources any more. Left for archives.A workflow for experimenting with block prompt injection. This is useful to cut down on embedding bleeding and also some useful ways to bypass blocks for ipadapter. We haven't really worked out what all blocks are useful for, but the ones we know about are marked. if you have consistent results on a block please share your findings so we can test and hopefully add them to the list.All Tips are welcome here &gt; <a target="_blank" rel="ugc" href="https://ko-fi.com/marduk191">https://ko-fi.com/marduk191</a>Known blocks:<ul><li>in8 = Subject matter</li><li>out0= Composition</li><li>out1= Style</li><li>out5- Details</li></ul>Repo instructions <a target="_blank" rel="ugc" href="https://github.com/DataCTE/prompt_injection">https://github.com/DataCTE/prompt_injection</a>There will be new nodes from matteo in a while so this might change significantly, yet work the same.**thread closedDiscord Support: <a target="_blank" rel="ugc" href="https://discord.gg/s3kj9VqpKc">https://discord.gg/s3kj9VqpKc</a>

marduk191's block prompt injection (archived)

V-ALPHA-2: initial upload for testers.

Files with SHA256: 70643873abc8b7860b0e370f84beaf46d71242e4c73b214610b5570c272d5489