Lets revisit what we can do with repomix.

Inside of a project:

npx repomix --init

repomix is a great tool that you can use to tell an LLM about all of the context of your project.

You can run it like this

npx repomix

And it will create a repomix-output.txt file that you can send to a LLM. You can drag the file over, or paste it in: (works best with claude)

cat repomix-output.txt | pbcopy

You can also pass this into a local llm, like

cat repomix-output.txt | ollama run gemma3:12b "Tell me what you think about all of this"

Check out more understanding-a-codebase

Tips

When you run repomix it shows you how many tokens the files use.

✔ Packing completed successfully!

📈 Top 5 Files by Character Count and Token Count:
──────────────────────────────────────────────────
1.  warc-viewer.py (19,802 chars, 4,023 tokens)
2.  tests/test_app.py (14,737 chars, 3,165 tokens)
3.  src/warc_viewer/app.py (14,513 chars, 2,859 tokens)
4.  src/warc_viewer/cdx.py (13,997 chars, 2,672 tokens)
5.  tests/test_db.py (13,766 chars, 3,141 tokens)

🔎 Security Check:
──────────────────
✔ No suspicious files detected.

📊 Pack Summary:
────────────────
  Total Files: 31 files
  Total Chars: 160,627 chars
 Total Tokens: 35,819 tokens
       Output: repomix-output.txt
     Security: ✔ No suspicious files detected

🎉 All Done!
Your repository has been successfully packed.

💡 Repomix is now available in your browser

Depending upon what’s in there, you can set up --ignore:

npx repomix --ignore "**/*.json,**/*.ipynb,**/*.mdx"

I have a bunch of mise tasks that depends on llm-dump which I define both in .config/mise/config.toml but then override in specific projects as needed.

[tasks.llm-dump]
description = 'Runs repomix'
run = ['repomix --ignore "output*,**/*.log,tmp/,inputs/**/*md,data_integrity/**/*.json"']