How do I install the Tooling First Automation skill?

Run npx skillmds@latest add selamy-labs/tooling-first-automation in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Tooling First Automation safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Tooling First Automation?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Tooling First Automation free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Tooling First Automation?

selamy-labs (@selamy-labs) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Tooling First Automation by selamy-labs

Name: Tooling First Automation
Author: selamy-labs

Tooling First Automation

Use when a task can be performed through existing callable tooling, MCP servers, browser automation, CLI wrappers, or service APIs before writing bespoke scripts or ad hoc automation. Especially relevant for browser/UI automation, external-system actions, and repeated operational workflows.

selamy-labs Updated Sep 17, 2026

File contents

Tooling-First Automation

Before writing new automation, inspect the tools already available in the environment and prefer the narrowest existing callable path that fits.

Workflow

Check the configured tool surface: MCP servers, built-in browser tools, repository scripts, CLIs, SDK helpers, and service-specific wrappers.
Use a first-party or official integration when it covers the task. For browser work, prefer a configured browser automation MCP or built-in browser tool before hand-writing Playwright, Puppeteer, Selenium, or CDP scripts.
If no existing tool fits, write the smallest scoped script needed and explain why the existing tools were insufficient.
If the right tool exists but is missing dependencies, broken, or unavailable in the current environment, fix or codify the tool installation/configuration in the owning source of truth before treating a one-off script as normal.
Verify the real tool path works with a representative action, not just that a package or config entry exists.

When Not To Use

The user explicitly asks for a from-scratch implementation.
The existing tool cannot perform the required action safely or with enough observability.
A short local script is the repository's established interface for the task.

Anti-Patterns

Writing bespoke browser automation while a browser MCP or built-in browser tool is configured and working.
Copying an API call into prose or shell snippets when a typed MCP/server tool already exposes the operation.
Declaring a tool "available" without launching it or completing a real action.
Leaving a dependency fix as an operator memory instead of codifying it in the source that provisions the environment.

---name: tooling-first-automationdescription: Use when a task can be performed through existing callable tooling, MCP servers, browser automation, CLI wrappers, or service APIs before writing bespoke scripts or ad hoc automation. Especially relevant for browser/UI automation, external-system actions, and repeated operational workflows.---# Tooling-First AutomationBefore writing new automation, inspect the tools already available in theenvironment and prefer the narrowest existing callable path that fits.## Workflow1. Check the configured tool surface: MCP servers, built-in browser tools, repository scripts, CLIs, SDK helpers, and service-specific wrappers.2. Use a first-party or official integration when it covers the task. For browser work, prefer a configured browser automation MCP or built-in browser tool before hand-writing Playwright, Puppeteer, Selenium, or CDP scripts.3. If no existing tool fits, write the smallest scoped script needed and explain why the existing tools were insufficient.4. If the right tool exists but is missing dependencies, broken, or unavailable in the current environment, fix or codify the tool installation/configuration in the owning source of truth before treating a one-off script as normal.5. Verify the real tool path works with a representative action, not just that a package or config entry exists.## When Not To Use- The user explicitly asks for a from-scratch implementation.- The existing tool cannot perform the required action safely or with enough observability.- A short local script is the repository's established interface for the task.## Anti-Patterns- Writing bespoke browser automation while a browser MCP or built-in browser tool is configured and working.- Copying an API call into prose or shell snippets when a typed MCP/server tool already exposes the operation.- Declaring a tool "available" without launching it or completing a real action.- Leaving a dependency fix as an operator memory instead of codifying it in the source that provisions the environment.

Frequently asked questions