No duplicate logic

Skill Fergius-Engineering/instincts/skills/no-duplicate-logic

Working-instinct skills for Claude Code: verify claims against the source, log for the bug you can't reproduce, write like a human, test so the test bites. A reflex layer on top of superpowers.

Install

npx -y skills add Fergius-Engineering/instincts --skill no-duplicate-logic

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when adding a case, check, or branch to a function that already exists, especially after jumping straight to the insertion point.

SKILL.md

2.0 KB, as published. Nobody here has run it

The rule

Before you add logic to a function that already exists, read the whole function. The case you're about to write is often already handled a few lines down. A second block that produces the same result doesn't replace the first — both run, and the output silently doubles. The fix you needed was usually three lines added to the block that's already there, not a new block beside it.

This is not the same as fixing in the shared layer. That's about where a bug lives across callers. This is narrower: don't paste a parallel branch inside one function without reading what's already in it.

Fires when

Adding a case, branch, check, or emit to an existing function, handler, switch, or validation pass — especially when you jumped straight to the insertion point without reading the rest of the function.

How to apply

Before adding logic that emits a result keyed by some id or name, search the whole function for that id or that result. If a block already handles it, extend that block instead of adding a new one. Read the function top to bottom before you insert anything into it.

Worked example

A validation function already emits a "bad-name" issue for a record. You add a new block to emit "bad-name" without noticing the existing one a screen above. Now every bad-named record produces two identical issues, and a count somewhere is quietly off by a factor of two. The real change was three lines inside the block that was already there. Reading the function before inserting would have shown it.

Red flags

Thought	Reality
"I'll just add the block here"	The block may already exist above; read first.
"It's a new case"	Check it's actually new before you write it.
"The function's long, I'll jump to the spot"	Jumping to the spot is how you duplicate the spot.