← 返回 Skills 市场
50
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install safe-change
功能描述
Map blast radius before shipping — run when editing a service, controller, hook, or shared utility to surface all importers, affected API routes, test gaps,...
安全使用建议
This skill appears to do what it says: read-only impact analysis plus an optional verify gate. Before installing/run it, ensure the agent is allowed to run shell commands in a safe environment (tests and npm scripts can execute arbitrary repository code and may have side effects). Review your project's test/build scripts for destructive actions or external network calls if you plan to run the verify gate via an agent. Also remember the scanner is regex-based and can undercount importers (barrels, dynamic imports, path aliases) — treat Low scores conservatively and manually inspect barrels/dynamic imports in high-risk areas.
功能分析
Type: OpenClaw Skill
Name: safe-change
Version: 1.0.1
The 'safe-change' skill is a utility designed to perform static impact analysis on TypeScript projects (NestJS/Next.js) before code changes are applied. The bundle consists of a Node.js script (scan-impact.mjs) that uses regex to identify importers, API routes, and environment variables, and a shell script (verify-gate.sh) that runs standard build and test commands. The instructions in SKILL.md and usage.md explicitly mandate a user checkpoint before any changes are made, and the scripts are primarily read-only or restricted to standard development workflows without evidence of data exfiltration, unauthorized execution, or malicious intent.
能力标签
能力评估
Purpose & Capability
The SKILL.md, README, and the included scripts implement an impact scanner and a verify gate. The scanner only reads project files (package.json and .ts/.tsx files) to build importers, routes, env var references, and recent migrations. The verify gate runs tsc/lint/test/build if present. All requested capabilities align with 'map blast radius' and 'verify build integrity'.
Instruction Scope
Instructions confine the agent to running the included scanner and verify script and to pausing for an explicit checkpoint. The scanner performs project-wide read-only file traversal and regex analysis (expected). The verify gate executes project commands (npm scripts, tsc) which may run arbitrary repo code (tests/build scripts can have side effects) — this is expected for a verify gate but worth noting as an operational risk.
Install Mechanism
No install spec; the skill is instruction + scripts included in the bundle. No external downloads or package installs are performed by the skill itself (scan-impact.mjs intentionally uses only Node stdlib).
Credentials
The skill declares no required environment variables or credentials. The scanner only locates process.env.* references in source files (it does not read those environment values). No unrelated secrets or external service credentials are requested.
Persistence & Privilege
The skill is not forced-always, does not modify other skills or global agent configuration, and contains no code that persistently enables itself. The verify gate runs local build/test commands but does not change other skills or system-wide settings.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install safe-change - 安装完成后，直接呼叫该 Skill 的名称或使用
/safe-change触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.1
- Added new banner asset: assets/banner.svg.
- Updated "When to Use" guidance in SKILL.md to clarify that most production incidents start as changes that felt small.
v1.0.0
Initial release of Safe Change — a skill to proactively map the blast radius of code changes and enforce a safer deploy process.
- Scans target files to surface importers, affected API routes, test gaps, ENV vars, and recent DB migrations.
- Assigns a risk score (Low/Medium/High) based on impact heuristics.
- Generates a detailed Markdown report for user review before proceeding.
- Includes a mandatory checkpoint requiring explicit user confirmation to continue.
- Provides a verify gate script to run type-check, lint, test, and build steps post-change.
- Designed to work with TypeScript, NestJS, Next.js; zero-dependency static analysis via regex.
元数据
常见问题
Safe change 是什么？
Map blast radius before shipping — run when editing a service, controller, hook, or shared utility to surface all importers, affected API routes, test gaps,... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 50 次。
如何安装 Safe change？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install safe-change」即可一键安装，无需额外配置。
Safe change 是免费的吗？
是的，Safe change 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Safe change 支持哪些平台？
Safe change 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Safe change？
由 brasco05（@brasco05）开发并维护，当前版本 v1.0.1。
推荐 Skills