lobotomized-claude-code

System-prompt overrides for Claude Code. Claude Code ships every model the same prompts — written for older Claudes and heavy with scaffolding, hedging, CAPS, and anti-laziness nagging that newer models don't need and often behave worse under. This repo cuts the bulk and rewrites the load-bearing parts in a register the model actually behaves better in.

skrabe/tweakcc-fixed reads these .md files and patches your installed Claude Code in place.

Which version do I use?

There are two prompt sets — pick the one matching your model:

Your model	Use this folder
Claude Opus 4.8	system-prompts-opus-4-8/ ← reshaped from scratch against the Opus 4.8 system card
Claude Opus 4.7 / 4.6	system-prompts-opus-4-7/

Not sure? Run claude --version and check which Opus you're on. The install below symlinks whichever set you choose; switching later is one ln -sfn.

What you get (Opus 4.8 set)

	Leaner by
Always-on prompts (inject every turn — harness, communication, doing-tasks, executing-actions, memory, core tools)	~30%
All behavior-shaping prompts	~36%
39 prompts removed entirely	unused features, duplicates, and safety theater

Less always-on prompt means a faster first response, more headroom before compaction, and less drift from contradictory rules — without losing anything the model relies on.

What it cuts vs. keeps

The Opus 4.8 set is shaped against the model's system card, not by taste:

• Cut: anti-laziness/"be thorough" scaffolding, CAPS theater (MUST/NEVER/ALWAYS), reflexive hedging and caveats, always-on upsells, restated rules, and feature docs you don't use — all things 4.8 either does by default or behaves worse under.
• Kept (and sharpened): the honesty contract (report failures truthfully, don't fake success), destructive-action confirmation, scope discipline, parallel-tool guidance, and exact output formats.
• Strengthened: on the one surface the card flags 4.8 as weaker — browser / computer-use / web-fetch — treat page and tool content as untrusted data, scrutinize intent, and confirm before destructive actions.

Install

git clone https://github.com/skrabe/lobotomized-claude-code ~/.tweakcc/lobotomized-claude-code

# pick your set (Opus 4.8 shown; use system-prompts-opus-4-7 for 4.7/4.6)
ln -sfn ~/.tweakcc/lobotomized-claude-code/system-prompts-opus-4-8 ~/.tweakcc/system-prompts
ln -sfn ~/.tweakcc/lobotomized-claude-code/system-reminders        ~/.tweakcc/system-reminders

# apply with the patcher
git clone https://github.com/skrabe/tweakcc-fixed ~/dev/tweakcc-fixed
cd ~/dev/tweakcc-fixed && pnpm install && pnpm build
node ~/dev/tweakcc-fixed/dist/index.mjs --apply

Re-run node ~/dev/tweakcc-fixed/dist/index.mjs --apply after each Claude Code update. To revert: node ~/dev/tweakcc-fixed/dist/index.mjs --restore.

Use skrabe/tweakcc-fixed (built from source, above) — not the tweakcc npm package. The skrabe fork carries the extraction + patching fixes these overrides depend on.

License

MIT. Derivative overrides of Anthropic's Claude Code prompts. These change model behavior in non-trivial ways and aren't endorsed by Anthropic — use on your local install at your own risk.