# Grade File

> Grade a file's responsibility hygiene

- Skill: `linardsliepenieks/grade-file` (Agent Skill)
- Install (CLI): `npx skillmds@latest add linardsliepenieks/grade-file`
- Raw SKILL.md: https://api.skillmd.com/api/skills/linardsliepenieks/grade-file/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: LinardsLiepenieks (https://skillmd.com/u/linardsliepenieks)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/linardsliepenieks/grade-file

---


# Grade a file's responsibility hygiene

Read-only. The user gives a **file** (default: the open one) and its **purpose** in one line. That purpose is the SRP ruler: anything that changes for a *different* reason is a finding. Do not edit; fixing is `simplify-code`'s job.

1. Read the file. Restate its purpose in one sentence; hold it as the ruler.
2. **Survey the neighbors**: `ls` the folder, grep for the same idea. A finding lands harder as "extract to *this existing seam*" or "duplicates *that* file" than in the abstract.
3. Name each deviation with a **Fowler smell** + `file:line`: *Divergent Change* (many reasons to change, the main hunt), *Bloaters* (Large Module, Long Method, Data Clumps), *Dispensables* (Dead Code, dead/stale comments, Duplication, Speculative Generality), *Couplers* (Feature Envy, Inappropriate Intimacy). Frame by cohesion (parts belong together?) + coupling (entangled with neighbors?).
4. Separate **dead** (remove) from **misplaced** (relocate). Accurate-but-wordy comments are not rot; say so if nothing is dead.

**Output** (prose): `**Score: N/10**` + the paradigm judged against · one line on what it gets right · findings grouped by smell with refs and the seam to extract to · an explicit dead-code/dead-docs line even if "none" · close with what would raise the score and to what.