Skip to content
Back to skills

Think Short Setup

ASecurity

Installs or removes terse-mode hooks. Explicit user invocation only.

  • 32 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added September 5, 2026
ai-agents

Security analysis

A100/100

Pro scans all 7 files and shows the line behind each finding

Scanned September 5, 2026

npx -y skills add kochetkov-ma/claude-brewcode --skill think-short-setup --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Think Short Setup?

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

Security grade badge for Think Short Setup
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/kochetkov-ma-think-short-setup/badge)](https://www.skillsdirectory.com/skills/kochetkov-ma-think-short-setup)

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: think-short-setup
description: "Installs or removes terse-mode hooks. Explicit user invocation only."
---

# Think-short hooks

## Resolve intent and target

1. Resolve exactly one canonical mode from `status`, `install`, `upgrade`, `enable`, `disable`, `uninstall`, `purge`, then project or personal scope. Show the exact target before mutation. With no mode given, resolve `status` when the assets are already present and `install` otherwise. `on`, `off`, `setup`, `remove`, `reset`, `create`, `update` and `cleanup` are not modes: read them as the canonical verb and echo the canonical name back.

## Modes

| Mode | Effect |
|------|--------|
| `status` | Report scope, registered entries, copied asset paths and their recorded version. Writes nothing. |
| `install` | Copy the two native scripts and the prompt described by `assets/INSTALL.md`, merge `SessionStart` and `UserPromptSubmit` entries by exact command string, and preserve unrelated hooks. |
| `upgrade` | Re-copy the same assets from the current plugin version and re-register any entry that went missing, restamping the recorded version. Keeps the parked-or-active state as it was. |
| `enable` | Restore parked assets by renaming each `.disabled` twin back to the filename the handler resolves. |
| `disable` | Park the copied assets by renaming them `.disabled`, leaving the bodies byte-identical, so the registered handlers no-op. |
| `uninstall` | Delete only the matching command entries and the three copied assets, plus any `.disabled` twin of them; remove empty directories only when owned by this workflow. |
| `purge` | `uninstall` plus removal of the workflow's own directory and any personal-scope override. State what will be deleted first. |

## Verify and report

2. Validate JSON, run both hook scripts with valid and malformed fixtures, and confirm a repeated `install`, `upgrade` or `uninstall` is idempotent.
3. Report the changed paths and require review through `/hooks`.

Handlers use one command string, timeout values in seconds, and no matcher for `UserPromptSubmit`. This Codex variant does not install a sub-agent prompt-rewrite hook.

Files in this skill

  • README.md139 B
  • SKILL.md2.1 KB
  • agents/openai.yaml259 B
  • assets/INSTALL.md854 B
  • assets/think-short-prompt-counter.mjs4.6 KB
  • assets/think-short-prompt.md634 B
  • assets/think-short-session.mjs4.7 KB

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.