← ClaudeAtlas

hands-onlisted

Load whenever the agent is about to tell the user to click, open, copy, fill in, configure or check something themselves ("go to Settings and...", "open the dashboard and...", "paste this in...", step-by-step instructions for the user). It makes the agent do every step it can do with its own tools (browser, CLI, API, files) and ask the user only for what it truly cannot or must not do (codes sent to their phone, passwords and secret keys, identity checks, payments, legal consent), all in one short message. Triggers on setup and configuration tasks, onboarding a new customer or account, connecting services (webhooks, apps, integrations), and on users who say "you do it", "stop sending me to do things", "why are you asking me if you can do it".
RabbitDesigns22/claude-skills · ★ 0 · AI & Automation · score 75
Install: claude install-skill RabbitDesigns22/claude-skills
# Hands on A user who hired an agent should not become the agent's hands. Every "now go to X and click Y" is work you pushed back to them, and they do it slower and with more mistakes than you would. The rule: **if you can do it, do it. Ask only for what you can't.** ## 1. Before writing any instruction for the user For each step of the plan, ask: **can I do this myself with the tools I have right now?** | The step | Who does it | |---|---| | Clicking, toggling, filling non-secret fields in a web dashboard where the user's session is already open in your browser | **You** | | Reading a value from a page, an API or a file (IDs, URLs, statuses, settings) | **You**. Never ask the user to copy an ID you can read | | Running a command, a test, an API read, a deploy to a preview or test environment | **You**. A deploy to a live app is public: ask first (see below) | | Writing or editing a file, a config, a document | **You**. Settings, permissions, hooks, and installing skills, plugins or packages: show the exact change and wait for the user's yes | | Checking that something worked (read it back, open the page, send the test) | **You** | | A one-time code sent to their phone or email (SMS, 2FA, verification) | **User** | | Passwords, API keys, tokens, secrets: typing or pasting them anywhere | **User** (you never handle the values) | | Identity checks, CAPTCHA, "confirm it's you" | **User** | | Payments, cards, anything that charges money | **User**, after a clear yes | | Accep