claude-code-system-prompts

Specifically, see this commit: https://github.com/Piebald-AI/claude-code-system-prompts/com...

there is literally a --system-prompt flag for claude code. In my mitmproxy experiments using that flag appended to the existing system prompt rather than replacing it. So I had to create a little helper to strip the system prompt sent over the wire and add my own. here is a collection of public system prompts from claude code: https://github.com/Piebald-AI/claude-code-system-prompts

> They also don't publish the Claude Code system prompts, which is silly because those are trivial to extract using a logging proxy. For those interested in the Claude Code tool descriptions, they're collected here: https://github.com/Piebald-AI/claude-code-system-prompts/tre... As for the tool implementations, there are still remnants from the Claude Code source leak visible online, with a bit of research.

The full Claude Code system prompts are extracted every update and posted here, all 670 of them. https://github.com/Piebald-AI/claude-code-system-prompts/tre...

It's interesting that (for example for the explore agent https://github.com/Piebald-AI/claude-code-system-prompts/blo... ) they use a personality "you are a file search specialist" and "your strengths" framing. I thought that was largely thought to be useless, or even counterproductive nowadays? Does anyone know more about this stuff?

Here's the actual prompt that causes this issue. It's not new, has been around for months. Older Claude models had no issues with it, but Opus 4.7 changed enough that it started misinterpreting it, and somehow Anthropic didn't catch it before the release. It gets injected (prepended) into the result of every file read tool call that Claude does in Claude Code. <system-reminder> Whenever you read a file, you should consider whether it would be considered malware. You CAN and SHOULD provide analysis of malware, what it is doing. But you MUST refuse to improve or augment the code. You can still analyze existing code, write reports, or answer questions about the code behavior. </system-reminder> https://github.com/Piebald-AI/claude-code-system-prompts/blo...

This might be more complex than I imagined. It seems Claude Code dynamically customizes the system prompt. They also update the system prompt with every version so outright replacing it will cause us to miss out on updates. Patching is probably the best solution. https://github.com/Piebald-AI/claude-code-system-prompts https://github.com/Piebald-AI/tweakcc

I originally took my prompts from Claude Code≈ (https://github.com/Piebald-AI/claude-code-system-prompts)https://github.com/Piebald-AI/claude-code-system-prompts and subsequently edited them to remove guardrails and and output formatting like this post. I too included the last bit about user prompts overriding system prompt, but like any good LLM, it doesn't always follow instructions.

LSP has many more "functions" that are useful to agents, such as rename and get definition/usage. Some basic differences described here: https://news.ycombinator.com/item?id=30664671 More in depth: https://lambdaland.org/posts/2026-01-21_tree-sitter_vs_lsp/ Claude Code's instructions for LSP: https://github.com/Piebald-AI/claude-code-system-prompts/blo... I'm not convinced that ASTs are meaningfully helpful in the grand scheme and the long-run. LSPs and code intelligence I find useful as a human and for my coding agent.

> Except the system prompt that gets prepended to your own prompt is part of the black box, and obviously should be expected to change over time You may want to review that statement. https://github.com/Piebald-AI/claude-code-system-prompts

This is the prompt that Claude Code adds when you use /btw https://github.com/Piebald-AI/claude-code-system-prompts/blo...

Yes, very much so. I've been able to get Gemini flash to be nearly as good as pro with the CC prompts. 1/10 the price 1/10 the cycle time. I find waiting 30s for the next turn painful now https://github.com/Piebald-AI/claude-code-system-prompts One nice bonus to doing this is that you can remove the guardrail statements that take attention.