SKIP TO CONTENT

Goalify

  • Plugin
  • Skill

Give your agent a big job and get proof back.

  • Questions asked once

    Every question comes first, once. The answers go into the brief.

  • Plan saved to file

    The run reads the brief file again, so it can start in a clean chat.

  • Proof it is done

    /goalify verify runs your tests again and writes VERIFIED or NOT VERIFIED.

Works with: Any agent that reads skills. Tested: Claude Code, Codex CLI, which both have /goal. Not tied to one model: plain Markdown and Python.

Pick your agent

npx skills add Aboudjem/goalify -g

Works in more than 70 agents: pick yours in the menu. Supported agents (new tab)Installs for all your projects. Only this project: drop -g.Its page on skills.sh (new tab)

Limit: A stopped run is not always done. Run verify.

github.com/Aboudjem/goalify (new tab)Version 3.1.4

What it does

  • Answer once, then leave. Every question comes first, so the run has no decision left to ask you.
  • Know when it is really done. The /goal line names the test command that must pass.
  • Keep the plan. The brief is a file that the run reads again and again.
  • See the plan first. A preview page shows every step and every check.
  • Get a receipt that can fail: make demo prints VERIFIED (2/2 checks passed). The --broken demo prints NOT VERIFIED (1/2 checks passed).

How it works

  1. 1. Homework. Goalify reads your project, the docs and your tests, and finds the commands that prove the job.
  2. 2. Questions, one time. It asks the few real choices together, then does not ask again.
  3. 3. Two things come out: the brief and the condition. A preview page also opens.
  4. 4. Type /clear, paste the /goal line and leave. The work starts in a clean chat.
  5. 5. Check it later: /goalify verify runs the checks again and writes a receipt.

Example

The line you paste looks like this. The path to the brief is inside it.

/goal Do everything in ~/acme/.goal/api-migration.md and prove it, done when the last turn quotes npm test passing and says ASYNC-OK. Stop after 40 turns.

Do not give /goal only a file path. The checker behind /goal cannot open files, so nothing would be checked.

More ways to install

Claude Code plugin

claude plugin marketplace add Aboudjem/10x
claude plugin install goalify@10x

Agent skill, for Cursor, Codex and other agents

npx skills add Aboudjem/goalify

Install by hand

git clone https://github.com/Aboudjem/goalify
cp -R goalify/skills/goalify ~/.claude/skills/

Limits

  • Goalify never does the job itself. It writes the brief and the condition, and the /goal command does the work.
  • A run that stops is not always finished. Run /goalify verify before you trust a green result.
  • /goal keeps your permission mode, so a run can still pause on a permission prompt.
  • It never guesses a cost in money or in tokens. Your step limit is your control.
Good for
A job too big to watch, in a project that has real tests.
Not for
A job that repeats without an end: Loopify is the tool for that.

Questions

What is Goalify?

Goalify is an open-source skill that prepares a long AI run. It writes a brief and a finish condition, and the run must show its checks passing.

How do I install Goalify in Claude Code?

Run claude plugin marketplace add Aboudjem/10x, then claude plugin install goalify@10x. Then type /goalify and describe the job; your Claude Code needs the /goal command.

Which agents does Goalify work with?

Tested on 17 November 2026: Claude Code and Codex CLI, the two agents with /goal. Install tested in Cursor, Gemini CLI, GitHub Copilot and OpenCode, where you start the run yourself.

Is Goalify free?

Yes, it is open source under the MIT license. It makes no network call of its own.