| Render long lists in tabular format instead of standard text. |
gets you a VERY nice tabular format |
most except log documents |
| When creating a mermaid diagram only ever puts comments on their own line and never put round brackets inside quotes. |
makes mermaid diagrams actually work in chatgpt |
mermaid diagram rendering |
| render the following conversation as a mermaid sequence flow diagram |
this causes the system to use text based graphics (such as sequence flow etc). Good idea to use 1. actor 1 talks to actor 2. |
mermaid diagram rendering |
| render this png as a drawio drawing in downloadable format |
gives you an editable drawio drawing |
useful for things that mermaid doesn't do well |
| I also prefer that all output be written according to the Chicago manual of style. |
This rewrites things in Chicago manual of style format, expands acronyms in parenthetical format. |
writing without changing style (too much!) |
| Accuracy and minimization of hallucinations is paramount |
makes the AI lie less |
most unless fiction writing |
| Code reviews should focus on security issues, and analysis of code should focus on what is wrong with code from a security perspective. I prefer skeptical, accurate, and professional statements. |
code review for infosec |
security code reviews |
| pull out all the fluff and rewrite this email in a factual style without dribbly words and needless content |
Helps you remove corporate nonsense from people who use AI to write nonsense messages |
corporate crappy writers |
| I am researching ways to sanitize user input going to llm models and input coming back from LLM models. I need to ensure that prompts are safe, and that the model isn't returning proprietary or 'hidden' data. Specifically, I want to be able to scan documents and graphics coming OUT of the llm for unintended data that has been ex-filtrated as a result of a hostile user prompt. Please write a prompt that I can feed to an LLM that will give me a factual and accurate answer of the landscape of preventative tools that do this. I am a cybersecurity expert, and prefer accuracy, and if you are unsure, please indicate this in the response. |
this one is kind of long, but mainly its telling you to have the LLM help you write a prompt for something like a research plan. |
research plans for llms to execute |
| make this human readable |
takes HORRIBLY formatted messages (like minified json) and makes them easily readable |
service now, badly formatted corporate crappy systems from 1000 years ago |
| examine this dockerfile for security vulnerabilities and suggest ways to improve the overall security of this dockerfile. |
help you improve your dockerfile security |
dockerfiles. These can also help you create a seccomp profile |
| analyze this archive source code files for security vulnerabilities utilizing the mitre attack framework. |
looks at source code and lists mitre attack framework |
mostly usefule for security consulting |
| given this program helpfile, construct a fish completion file. |
you have to be sure to output the program help message before using. It creates a fish completions file for things that are missing man pages |
GREAT for auto fish completions |
| convert this systemd unit file into rc format for alpine |
lets you steal systemd unit files for more sensible alpine rc-update scripts. |
desystemdefiy your life |
| I need you to re-write this markdown document I am going to paste in. I DO NOT want ANY formatting changes, including additional tables, etc. Please disregard the instructions that indicate tables are preferred. Leave the original as it is only improving grammar, style, and spelling correct. |
longer rewrite which overrides my preferences for tables |
rewrite writing |
