Bug Discovery Code Assistant — ChatGPT & Claude.ai prompt: find bugs
Turns a chat model into a code reviewer: it hunts logical, syntax and runtime errors and suggests fixes. Handy when you need a second pass over an unfamiliar file.
Open on GitHubWhat Bug Discovery Code Assistant does
The prompt puts the model in the role of a software expert with an eye for bugs and inefficiencies. It walks the code you paste, flags logical, syntax and runtime errors, then proposes fixes or improvements. Performance and security are both called out as things to check.
There is no script behind it, just prompt text. A ${code} placeholder keeps the template reusable, so you swap in a new snippet instead of rewriting the instructions each time. Feedback is meant to stay short and to the point.
It runs in ChatGPT and Claude.ai, meaning a plain chat with no access to your repository. You paste the code yourself, and on a large file the model only sees what you sent. If the bug depends on modules you left out, the review will miss it — worth checking that every related piece is in the message.
What it helps with
- spot a logic error in someone else's code before merging
- check a snippet for security holes and slow spots
- get a list of fixes for an unfamiliar file
- reuse one review template across projects
Example prompts for the agent
- Review this code and find the bugs: ${code}
- Check this snippet for logical and runtime errors and suggest fixes: ${code}
- Look at this code for security and performance problems: ${code}
How to install Bug Discovery Code Assistant
How to use
Paste the text at the start of a chat or into project instructions
Manually: put the files in place
.claude/commands/prompts.csv.mdQuestions about Bug Discovery Code Assistant
How do I install Bug Discovery Code Assistant in Claude?
There is nothing to install: paste the prompt into the chat as a normal message. Copy the text, replace ${code} with your snippet and send it.
What does Bug Discovery Code Assistant do?
It reads the code you send and looks for logical, syntax and runtime errors, then suggests fixes. It also checks performance and security.
How is it different from just asking the model to find bugs?
It sets a role and a fixed sequence: full review of the code, error hunt, proposed fixes. The ${code} placeholder lets you reuse the template without rewriting it.
Source text
prompts.csv · CC0-1.0
Act as a Bug Discovery Code Assistant. You are an expert in software development with a keen eye for spotting bugs and inefficiencies.
Your task is to analyze code and identify potential bugs or issues.
You will:
- Review the provided code thoroughly
- Identify any logical, syntax, or runtime errors
- Suggest possible fixes or improvements
Rules:
- Focus on both performance and security aspects
- Provide clear, concise feedback
- Use variable placeholders (e.g., ${code}) to make the prompt reusable
© weiruo-c · CC0-1.0 · f/prompts.chat
