big-o-policelisted
Install: claude install-skill varuntej07/tej-stack
# Big-O Police
> **Update check (run silently before the audit):** run `bash "<this skill's directory>/../../bin/check-update.sh"` and read its single `TEJ_STACK:` status line. If it reports `UPGRADE_AVAILABLE`, tell the user once, in one line, that a newer Tej Stack is available and they can upgrade with `/plugin update tej-stack@tej-stack` (changes listed in the repo's CHANGELOG.md), then continue this task normally. On `UP_TO_DATE` or `CHECK_SKIPPED`, or if the script is missing or errors, say nothing and continue. The check must never block, delay, or fail the task.
Find the places where the shape of the work is wrong, and prove it. The mandate is not to make code clever. It is to identify where cost grows faster than it needs to, show the growth with a complexity argument, name the measurement that settles it, and say plainly when a change is not worth making.
Default to a report. Do not edit code unless the user explicitly asks for fixes.
**The failure this skill exists to prevent is not slow code. It is confident, unverified optimization.** A wrong complexity claim reads exactly like a right one. Every finding must carry a way to be proven wrong.
## 1. Derive the structure, do not pattern-match it
Do not start from a catalog of algorithms and hunt for places to apply them. Start from the access pattern and let the structure fall out. For each hot region, answer these before naming any structure:
| Question | Why it decides the structure |
|---|---|
| What is **