define-goal
Help the user define a concrete, measurable goal before starting work, especially when they ask to use the goal tool, create a goal, set an objective, clarify success criteria, or turn a fuzzy intention into a quantitative outcome. Use this skill for goal creation and goal refinement only; it does not manage durable snapshots, decision logs, or long-running execution artifacts.
Security Assessment
About define-goal
Define Goal is a lightweight reasoning-discipline skill that helps an agent turn a fuzzy user intention into a concrete, measurable objective before any implementation work begins. The problem it addresses is that agents often start executing against vague requests like "make checkout faster" or "keep investigating," which have no stopping condition and lead to aimless effort. This skill forces the objective to name a specific outcome, the artifact or system involved, how completion will be verified, and what is in and out of scope.
The workflow first confirms that goal definition is actually needed, then restates the likely goal in concrete terms, makes it quantitative where the domain supports it (pass/fail validators, quality thresholds like latency or error rate, artifact constraints, evidence counts), and repairs weak or pure-activity goals before setting them. It checks existing goal state via a `get_goal` call and only creates a goal with `create_goal` once the objective passes an explicit quality bar that answers what will be true, what evidence proves it, what threshold defines success, and what should cause the agent to stop and ask. The skill deliberately excludes durable snapshots, decision logs, ledgers, and resume files.
It is aimed at agent builders and users of the Codex/OpenAI goal tooling who want objective-driven execution, and at anyone who wants an agent to negotiate a crisp definition of done rather than grinding. Typical use cases include scoping bug fixes as reproduce-then-fix objectives, bounding PR review-comment resolution, and converting performance requests into benchmarked targets.
FAQ
When should this skill be used?
When the user explicitly asks to create or set a goal, requests the goal tool, or wants help turning a fuzzy intention into a measurable objective. For ordinary implementation work it says to do the work directly rather than forcing goal creation.
What tools does it rely on?
It calls `get_goal` to check active goal state and `create_goal` to record an objective, but only after the objective meets its quality bar.
What makes a goal acceptable to this skill?
The objective must name a concrete outcome, the evidence that proves it, a quantitative or binary success threshold, relevant scope boundaries, and a stop condition for asking the user. Pure activity goals like 'make progress' are rejected unless sharpened.
Does it manage long-running artifacts or planning files?
No. It is explicitly scoped to goal definition and goal-tool creation only, and does not create intermediate planning artifacts, durable snapshots, ledgers, decision logs, or resume files.
Install define-goal
Quick Setup:
- Copy the skill folder to
.claude/skills/ - Claude will automatically detect and use the skill
Repository
openai/skills