agentsclimarketplace

Data loss guard

Skill pipipip169/fable5-handoff/sources/adamentwistle-fable-skills/data-loss-guard

Stop-and-check protocol before any destructive or hard-to-reverse operation — rm, overwrite, force-push, reset --hard, DROP, bulk delete, killing processes. Use the moment a planned action would destroy state that cannot be trivially recreated.From its SKILL.md

Install
npx -y skills add pipipip169/fable5-handoff --skill data-loss-guard

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

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 1 stars1 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.

SKILL.md

1.9 KB, 377 tokens by cl100k_base, as published. Nobody here has run it

data-loss-guard

Almost every catastrophic agent failure is a destructive command run on a wrong assumption. Before any destructive action:

  1. Look at the target first. ls the directory before rm -rf; git status/git stash list before reset --hard or checkout .; SELECT before DELETE/UPDATE (same WHERE clause, check the row count). If what you find contradicts what you expected — files you didn't create, more rows than predicted — stop and surface it instead of proceeding.
  2. Distinguish recoverable from gone. Committed work is recoverable via reflog; uncommitted work is not. A dropped column is gone. Overwritten files have no undo. Rank the action's blast radius honestly before choosing convenience.
  3. Prefer the reversible variant: git stash over discard; move to a trash/backup path over rm; soft-delete or export-then-delete over hard delete; --dry-run flags wherever they exist (rsync, terraform, kubectl, many CLIs) on the first pass.
  4. Scope the command tightly. No wildcards where an explicit list will do; absolute paths so a surprise cwd can't redirect the deletion; never rm -rf a variable-built path without checking the variable is non-empty and expected.
  5. Uncommitted user work is sacred. If the working tree holds changes you didn't make, don't clean, checkout, or stash over them without the user's explicit say-so — that's their morning, not your scratch space.
  6. Force-push and history rewrites on shared branches need explicit user authorization, every time.

Rule of thumb: cheap check, irreversible mistake — the one-command look-before-delete is always worth it.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.