Hey there,

I have both of the big Claude Code process plugins installed in the same repo. It took me a while to notice that six of their skills do the exact same job.

Nobody warns you about that, and the thing that decides which one wins is smaller than you would expect.

Six Claude Code skills collide, and one hook decides them

Superpowers ships 14 skills. Compound Engineering ships 33. Both cover brainstorming, planning, execution, debugging, code review, and worktrees – so once both are installed, every one of those is a quiet fork in the road. What settles it is not quality. Superpowers ships a SessionStart hook and Compound Engineering ships none, so Superpowers is already in the room before you type anything and its version wins by default. A Hacker News thread claims Superpowers is token-hungry, so I measured both installs rather than reading the READMEs: the always-on injection is 777 tokens, about 0.4% of a 200k window, while subagent-driven-development runs about 8,084 once it loads and Compound Engineering's heaviest is 3,650. Which makes the usual framing backwards – one is cheap to install and expensive every time it fires, the other is free until you call it. The fix for the collisions turned out to be one line in AGENTS.md.

Read the measured comparison →

What compound engineering is, and the file that makes it compound

Rebuilt this week against the current plugin: 33 skills, no standalone agents, and the numbers now match what the installed version actually contains. The idea underneath it is that every task should leave the codebase easier to work in than it found it, which only happens if the process lives somewhere Claude reads first. That somewhere is AGENTS.md, a sibling to CLAUDE.md that loads ahead of it, and the post walks through a real one – the pipeline it encodes, the tier rules that scale process to risk, and why splitting process out of CLAUDE.md is a read-order decision rather than a tidiness one. The section most people skip is the one that matters: without the compounding step, the plugin degrades into a todo runner.

See the AGENTS.md walkthrough →

/code-review vs /ce-code-review, when each one wins

The third leg of the same cluster, and the clearest case of two skills that look like duplicates and are not. Claude Code ships /code-review and /simplify natively; the plugin adds /ce-code-review and /ce-simplify-code on top. The post is the side-by-side: the quick-review short-circuit most people miss, a decision matrix for which to reach for, and the cost gotcha worth knowing – the plugin forces Sonnet on most personas, so skipping that override quietly runs a review at three to four times the price.

Read the comparison →

Reply and tell me which one was the read for you. I read every reply.

– Alex

P.S. – The conversations behind reads like these happen in the WotAI Skool community. 760+ builders, three live calls a week, free to jump in.