Etc Skill — Development skill for Claude Code
An agent skill for "fix these two, then do the rest the same way" — derive the rule, sweep the whole scope, report the rule instead of a file count.
How to install Etc Skill
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open vladmoseev/etc-skill and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Etc Skill does
An agent skill for "fix these two, then do the rest the same way" — derive the rule, sweep the whole scope, report the rule instead of a file count.
Alternatives in Development
- Check A11y — Report WCAG contrast for the theme and the minimal fix to reach AA 452 ★
- Dashclaw Quality — Recurring find-and-fix quality pass over the whole DashClaw app — browser smoke (frontend-verify) + code gates 297 ★
- Fix Job — Diagnose and fix a failed VASP job, then restart it 155 ★
README
etc — an agent skill
"Fix items 1 and 2, then do the rest the same way."
You show a coding agent two or three corrections as examples. It fixes exactly those two or three, reports "done," and hands back a file that still has twenty instances of the thing you just complained about.
This skill fixes that. It turns "and so on" into a working instruction.
It ships as a `SKILL.md` because that's a format some agents load by themselves, but nothing in it is tied to one of them: it's a page of plain Markdown describing a behaviour, and it works in any agent you can give standing instructions to. See [Install](#install).
What it does
When the skill fires, the agent stops treating your examples as a task list and starts treating them as evidence of a rule. Four steps:
- State the rule in words before editing anything. "You fixed
XandY. The rule I derived: dates useYYYY-MM-DD. It covers 23 more places." A rule that isn't stated doesn't exist — and a stated rule lets you catch a misreading in one line instead of after twenty wrong edits. - Search the whole stated scope, not just the convenient part, and show the full list of hits.
- Sort the hits into three piles — clearly covered, covered-but-different (fix and name the difference), and looks-similar-but-isn't (leave alone and say why). The third pile is the important one: it's the proof the agent understood the rule instead of grepping for lookalikes.
- Report the rule, not the count. "Fixed 23 files" tells you nothing. "Rule: X. 23 places, all fixed. Three left alone because Y" tells you whether to trust the pass.
It also covers the failure modes around the edges: don't stall waiting for the obvious rule to be confirmed, don't let the rule grow into unrelated cleanup, re-check the examples you were shown (people often fix them only partway), and treat zero findings as a legitimate answer.
When it fires
On any phrase that hands over a pattern inste
Related Skills
Evil Mode
EVIL mode — adversarial sweep with scored findings, adversarial verification, and loop-until-dry. The upgrade
Se Plan
Generate the implementation/integration plan — derive the CR build order from the graph's depends-on DAG, cut
Autofix Issues
Batch — triage and fix ALL eligible open issues, then produce one comprehensive report.
Editscore Ladder Run
Run an EditScore-7B pairwise ladder for an image-edit generator (SDEdit / diffusion / LoRA / distilled step-co
Logic Health
Sweep a whole codebase or directory for logic correctness — use before a release or to identify risk hotspots
Sweep
Daily end-of-day Gmail inbox sweep — auto-label the obvious, flag the rest for review, log decisions for futur
Related Agents
Roblox Auditor
Audits a whole Roblox project and returns a scored report instead of file contents. Use when the user asks to
Hydra-sentinel-scan — Fast Integration Sweep
You are the first line of defense against integration breakage. You run AFTER code changes and perform a fast
Code Quality Author
Use to FIX/refactor code to meet the quality standards — resolve clean-code, DRY, KISS violations with the sma