Safely runs registered repository scripts by following a registry and requiring approval for high-impact actions.

Install

mkdir -p .claude/skills/ai-scripts && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/19337" && unzip -o skill.zip -d .claude/skills/ai-scripts && rm skill.zip

Installs to .claude/skills/ai-scripts

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Use registered scripts with risk-based approvals and evidence.
62 charsno explicit “when” trigger
Intermediate

Key capabilities

  • Execute registered repository scripts
  • Perform discovery and evidence gathering
  • Apply change-aware context to operations
  • Verify repository wiring and structure
  • Gate high-impact script execution with approvals

How it works

It enforces a strict registry-based selection process for scripts and mandates approval for any mutation-heavy or high-impact operations.

Inputs & outputs

You give it
Script name and target query
You get back
Execution results or file previews

When to use ai-scripts

  • Search for specific patterns in the codebase
  • Verify repository structure or wiring
  • Safely execute high-impact scripts
  • Retrieve code diffs and file previews

About this skill

AI Scripts Skill

Use this skill when selecting or running repository scripts from scripts/ai/.

Selection order

  1. Confirm script is listed in docs/ai/script-registry.json.
  2. Confirm doc parity in docs/ai/script-registry.md and docs/ai/scripts-reference.md.
  3. Prefer the narrowest read-only wrapper that answers the request.

Preferred wrappers

  • Discovery/evidence: ai-search, preview-file, query-usage
  • Change-aware context: ai-diff-context, git-forensics
  • Wiring checks: ai-doc-check, ai-search doctor
  • Verification: ai-verify

Approval gate

Ask before mutating or high-impact wrappers, including ai-edit, ai-rollback, install-mandatory-tools, and broad context packers.

Evidence commands

git status --short
AI_OUTPUT=json bash scripts/ai/ai-search.sh changed-text <query> . --fixed
AI_OUTPUT=json bash scripts/ai/ai-search.sh staged-text <query> . --fixed
AI_OUTPUT=json bash scripts/ai/ai-search.sh tracked <query> . --fixed

Use preview-file for line-level context:

AI_OUTPUT=json bash scripts/ai/preview-file.sh <path> --around <line> --context 30

If runtime lacks bash or php, record the command as not run and keep claims bounded.

When not to use it

  • When scripts are not listed in the registry
  • When the runtime lacks bash or php

Prerequisites

bashphp

Limitations

  • Requires script registration in docs/ai/script-registry.json
  • Claims must be bounded if runtime lacks bash or php

How it compares

It replaces arbitrary script execution with a governed, evidence-based workflow that requires registry parity and explicit approval for changes.

Compared to similar skills

ai-scripts side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
ai-scripts (this skill)019dReviewIntermediate
run-nx-generator52moReviewIntermediate
upgrading-expo33moReviewIntermediate
bump-go-dependencies12moReviewIntermediate

Try saying

Example prompts that trigger this skill in your AI assistant.

Search skills

Search the agent skills registry