Does this skill block read operations?

No, all read-only operations, searches, and local file modifications are allowed to proceed without confirmation.

Are Linear operations protected by this workflow?

By default, Linear writes are considered internal tracking and are allowed without confirmation to maintain developer velocity.

Which systems are covered by this safety skill?

The skill monitors and enforces safety checks for Jira, Confluence, BitBucket, and Slack write operations.

Why does the skill reject responses like 'ok' or 'sure'?

The skill requires explicit confirmation such as 'yes' or 'proceed' to eliminate ambiguity and prevent accidental triggers from casual conversation.

Does it support batch updates for multiple tickets?

Yes, it can handle batch operations by showing a summary of all intended changes and allowing for a single bulk confirmation or granular per-item approval.

External System Safety Guardrails

Name: External System Safety Guardrails
Author: duongdev

byduongdev

•

セキュリティとテスト

Enforces mandatory confirmation workflows and content previews for all write operations to external platforms like Jira, Confluence, BitBucket, and Slack.

This skill acts as a critical safety layer for Claude Code, preventing accidental or unauthorized modifications to external project management and communication tools. It automatically detects potential write operations—such as updating Jira tickets, creating Confluence pages, or posting Slack messages—and forces a stop-and-confirm workflow. By presenting a detailed preview of the intended change and requiring an explicit 'yes' confirmation, it ensures high-stakes external updates are deliberate, accurate, and auditable, while still allowing seamless read-only access and local development tasks.

主な機能

01Mandatory content previews showing exactly what will be posted or modified

02Comprehensive audit logging of all confirmed external operations

035 GitHub stars

04Automatic write detection for Jira, Confluence, BitBucket, and Slack

05Strict confirmation logic requiring explicit responses to prevent accidental execution

06Granular batch operation support with per-item confirmation options

ユースケース

01Safe-guarding Confluence documentation from unintended overwrites or deletions

02Preventing accidental status changes or comment posts in production Jira environments

03Reviewing Slack notifications or team-wide announcements before they are broadcast

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add duongdev/ccpm external-system-safety

For use in Claude.ai and ChatGPT

Download Skill