Skip to content
Back to skills

Verify Repository Change

ASecurity

Validate one planned repository change with bounded allowlisted commands and scope checks before committing. Use for Solve test, lint, typecheck, build, changed-file validation, command safety, and incremental commit phases.

  • 85 stars
  • 0 votes
  • 0 copies
  • 4 views
  • Added September 3, 2026
devopsrustshell

Works with

  • cli

Security analysis

A100/100

Pro scans all 2 files and shows the line behind each finding

Scanned September 3, 2026

npx -y skills add elixpo/elixpo_chapter --skill verify-repository-change --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Verify Repository Change?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Verify Repository Change
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/elixpo-verify-repository-change/badge)](https://www.skillsdirectory.com/skills/elixpo-verify-repository-change)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: verify-repository-change
description: Validate one planned repository change with bounded allowlisted commands and scope checks before committing. Use for Solve test, lint, typecheck, build, changed-file validation, command safety, and incremental commit phases.
---

# Verify Repository Change

Confirm that every changed path belongs to the current step and full plan. Reject
untracked or modified scope outside those paths, symlinks, empty changes, and
generated surprises.

Parse verification commands into argument vectors and execute without a shell.
Require a configured prefix allowlist. Reject composition operators, redirects,
installers, network clients, deployment, publishing, destructive commands, and
credential access. Apply per-command timeouts and RTK output compression.

Strip token, key, secret, password, and private-key environment variables from
every target-repository subprocess. Permit at most one configured dependency
setup command. Treat setup as untrusted code in the disposable sandbox and never
record its output as a successful behavioral check.

Continue through the bounded command plan after a non-zero exit. Do not retry
automatically, alter the command, or weaken configuration. Preserve compact
failure output as a verification exception for diagnosis and PR disclosure.
Treat failed checks as advisory only after the requested implementation is
complete and self-reviewed.

After all checks are attempted, verify they introduced no tracked changes, then
stage only declared files. Require a conventional
commit message and a non-empty staged diff, then create exactly one local commit.
Require a clean working tree after the final step. Do not push.

Files in this skill

  • SKILL.md1.7 KB
  • agents/openai.yaml222 B

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.