How do I install the Goal Command skill?

Run npx skillmds@latest add uitbreidenos/goal-command 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 Goal Command safe to use?

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

Which AI agents work with Goal Command?

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 Goal Command free to use?

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

Who published Goal Command?

UitbreidenOS (@uitbreidenos) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Goal Command by uitbreidenos — Agent Skill

Name: Goal Command
Rating: 4.2 (5 reviews)
Author: UitbreidenOS

File contents

/goal — Autonomous Task Completion

When to activate

User wants Claude to keep working on a task without stopping to check in; user wants to set a completion condition and walk away; user asks about autonomous operation or keeping Claude running until a specific outcome is reached.

When NOT to use

Simple single-step tasks where one response is sufficient; tasks where the user wants Claude to pause and confirm after every action; interactive debugging sessions where back-and-forth is the point.

Instructions

Syntax:

/goal <completion condition>

The condition is evaluated after every assistant turn. Claude keeps working — writing code, running commands, seeing failures, adjusting — until the condition holds, then stops and reports.

Writing good conditions:

Natural language works. The condition should be observable and unambiguous:

All tests pass — Claude runs the test suite, fixes failures, reruns, until green
The PR is created — Claude finishes the work and opens a PR
The migration runs without errors — Claude applies the migration, checks for errors, fixes schema issues
tsc --noEmit exits 0 — Claude resolves TypeScript errors until the compiler is clean
CHANGELOG.md exists and has today's date — Claude writes the changelog file

Bad conditions to avoid:

Subjective: "looks good", "is clean" — not verifiable by Claude
Open-ended: "keep improving the code" — no stopping condition
Time-based: "run for one hour" — not an outcome

Combine with effort level for maximum autonomy:

/goal All tests pass
/effort xhigh

Interrupting: Send any message to interrupt, or delete .claude/goal to cancel. The goal state persists across context compactions — Claude remembers the goal even after the context window compresses.

Background sessions: Works with claude --bg. Set the goal, close your terminal, come back when it's done.

What happens in each turn:

Claude takes actions (edits files, runs commands)
Evaluates: does the condition hold?
If no — continues
If yes — stops and reports what was done

Example

/goal All TypeScript errors are resolved and tsc --noEmit exits 0

Claude runs tsc --noEmit, reads the error list, fixes each error, runs again, sees remaining errors, fixes those, runs again — loop continues until zero errors. Then stops and reports: "Resolved 14 TypeScript errors across 6 files. tsc --noEmit exits clean."

---name: goal-commanddescription: User wants Claude to keep working on a task without stopping to check in; user wants to set a completion condition and walk away; user asks about a...---# /goal — Autonomous Task Completion## When to activateUser wants Claude to keep working on a task without stopping to check in; user wants to set a completion condition and walk away; user asks about autonomous operation or keeping Claude running until a specific outcome is reached.## When NOT to useSimple single-step tasks where one response is sufficient; tasks where the user wants Claude to pause and confirm after every action; interactive debugging sessions where back-and-forth is the point.## Instructions**Syntax:**```/goal <completion condition>```The condition is evaluated after every assistant turn. Claude keeps working — writing code, running commands, seeing failures, adjusting — until the condition holds, then stops and reports.**Writing good conditions:**Natural language works. The condition should be observable and unambiguous:- `All tests pass` — Claude runs the test suite, fixes failures, reruns, until green- `The PR is created` — Claude finishes the work and opens a PR- `The migration runs without errors` — Claude applies the migration, checks for errors, fixes schema issues- `tsc --noEmit exits 0` — Claude resolves TypeScript errors until the compiler is clean- `CHANGELOG.md exists and has today's date` — Claude writes the changelog file**Bad conditions to avoid:**- Subjective: "looks good", "is clean" — not verifiable by Claude- Open-ended: "keep improving the code" — no stopping condition- Time-based: "run for one hour" — not an outcome**Combine with effort level** for maximum autonomy:```/goal All tests pass/effort xhigh```**Interrupting:** Send any message to interrupt, or delete `.claude/goal` to cancel. The goal state persists across context compactions — Claude remembers the goal even after the context window compresses.**Background sessions:** Works with `claude --bg`. Set the goal, close your terminal, come back when it's done.**What happens in each turn:**1. Claude takes actions (edits files, runs commands)2. Evaluates: does the condition hold?3. If no — continues4. If yes — stops and reports what was done## Example```/goal All TypeScript errors are resolved and tsc --noEmit exits 0```Claude runs `tsc --noEmit`, reads the error list, fixes each error, runs again, sees remaining errors, fixes those, runs again — loop continues until zero errors. Then stops and reports: "Resolved 14 TypeScript errors across 6 files. `tsc --noEmit` exits clean."---

Frequently asked questions

Goal Command