20 ready messages

Messages for your agent

Copy, fill in the [brackets], paste. Each one says when to use it and what to expect back.

  1. 01

    Start a project

    When: You have an idea and an empty session.

    I want to start a new project. Help me create its folder in my Development folder, then interview me one question at a time to write SPEC.md.

    Expect: A folder name suggestion, then questions about people, problems and goals, never code.

  2. 02

    Write the spec

    When: You've talked about the idea and want it written down.

    Write SPEC.md from what we've discussed: a plain-language summary first (what, who for, must-haves, what's out, how we'll know it works), then the technical sections.

    Expect: A spec you can read in two minutes, and a request for your OK.

  3. 03

    Change the spec

    When: You changed your mind about a feature.

    I've changed my mind: [describe the change]. Update SPEC.md and the plan first, show me what changed, then wait for my OK before touching code.

    Expect: The spec and plan updated first; code only after you agree.

  4. 04

    Add logins

    When: People need their own accounts.

    We need logins. Update the spec first, then add Better Auth following the better-auth best-practices and security skills, with rate limits on sign-in. Explain each step.

    Expect: A spec update, then a branch with logins, a check of every endpoint, and a Preview.

  5. 05

    Add a database

    When: Your app needs to remember things.

    Our app needs to store data. Update the spec, then guide me through connecting Neon from Vercel, and use a development branch on my computer, never production.

    Expect: Click-by-click steps for you in Vercel; the code part done by your agent.

  6. 06

    Add AI

    When: You want an AI feature in your app.

    Add an AI feature: [describe it]. Calls on the server only, Tokentide first with the Vercel AI Gateway as fallback, a token limit on every request, and no personal data in prompts.

    Expect: A plan for the feature, the keys you'll need to add yourself, and a fallback test.

  7. 07

    Explain what you just did

    When: Your agent did something you didn't follow.

    Explain what you just did in three plain sentences: what changed, why, and how I can see it.

    Expect: A short, jargon-free explanation, with each new term explained.

  8. 08

    Something broke

    When: The app doesn't do what it should.

    Something broke. What I did: [steps]. What I expected: [expected]. What happened: [what you saw, plus the exact error]. Find the root cause, fix it, and add a test so it can't come back.

    Expect: Your agent reproduces it, explains the cause, fixes the root and proves it with a test.

  9. 09

    Go back to the last version that worked

    When: You just want it working again.

    Go back to the last version that worked, without deleting history. Then tell me what changed since.

    Expect: A working app again, using git revert or a new branch.

  10. 10

    Security review

    When: Before publishing, or after adding logins, data or AI.

    Do a security review with the security-and-hardening skill: secrets, endpoints, input validation, rate limits and headers. Report each finding in plain words, most serious first.

    Expect: A ranked list of findings, each with a suggested fix.

  11. 11

    Quality bar review

    When: Right before you publish.

    Check the project against the Agentic Makers quality bar (works, safe, handles errors, fast, accessible and responsive, private, findable) and report pass or fail per item, in plain words.

    Expect: A pass/fail list and a fix for anything that fails.

  12. 12

    Publish

    When: The Preview looks right and the review passed.

    Publish: merge the pull request into main, wait for Vercel, open the live address and share it with me. Update tasks/todo.md.

    Expect: Your live address, and the checklist updated.

  13. 13

    Resume a session

    When: You're back after a break.

    Read SPEC.md and tasks/todo.md, pull any changes, and tell me where we are and what's next.

    Expect: A short status: the Path stage, what's done and the next task.

  14. 14

    Roll back the live site

    When: The live site broke after a publish.

    The live site is broken. Guide me through Vercel's Instant Rollback to the last good deployment, then find the cause from the runtime logs.

    Expect: Click-by-click rollback steps, then a fix on a branch with a Preview.

  15. 15

    A key leaked

    When: A key ended up in the chat, a file or a commit.

    A key may have leaked and I've revoked it. Check no file or commit contains it, and tell me where to add the new one. Don't show any value.

    Expect: A check of files and history, and where the new key goes.

  16. 16

    Where are we on the Path?

    When: You want the big picture.

    Show me where we are on the Path (Setup → New project → Spec → Plan → Build & preview → Publish) and what's left before our first live app.

    Expect: The six stages with yours highlighted.

  17. 17

    Show me the history

    When: You want to see what's been done.

    Show me my project's commits and explain each one in a sentence.

    Expect: Your project's story, one line per step.

  18. 18

    Check my keys (safely)

    When: Something says a key is missing.

    Check which environment variables are set or missing, locally and in Vercel. Names only, never values.

    Expect: A list of names with “set” or “missing”.

  19. 19

    Slow down, I'm lost

    When: It's going too fast.

    I'm lost. Slow down, explain where we are in plain words, and offer me a simpler way forward.

    Expect: A calmer pace and a simpler explanation.

  20. 20

    Teach me this

    When: You want to understand an idea, not just get it done.

    Teach me what [term] means, with an everyday analogy and an example from my own project. Then ask me one quick question to check I got it.

    Expect: A short lesson and one quick check.

Tested with Claude Code before launch · Last reviewed Nov 8, 2026