Document inline vs reviewable /goal output shapes

[sungjunlee]

Problem

skills/craft-prompt/references/goal-conditions.md currently focuses on a single inline /goal condition string. That is correct for small goals, but recent practice shows a second useful shape: a reviewable goal spec saved as a markdown artifact before activation.

The distinction matters because long-running goals can burn substantial time and tokens if the contract is wrong.

Proposed change

Add a short section to skills/craft-prompt/references/goal-conditions.md that distinguishes two output shapes:

1. Inline condition: a short 1-3 sentence /goal <condition> for small, clear tasks.
2. Reviewable goal spec: a markdown contract for expensive, ambiguous, or multi-hour work, optionally saved under goals/<slug>.md and activated only after review.

The reviewable spec should make clear that the file is not active goal state. It becomes active only when the user runs something like /goal @goals/<slug>.md, where supported.

Acceptance criteria

• The reference explains when to use inline vs file-backed goal specs.
• The reviewable spec template includes outcome, evidence, constraints, non-goals, scope, budget, blocked stop condition, and final report expectations.
• Claude/Codex differences are explicit where they affect behavior, especially transcript-visible evidence for evaluator judgment.
• The guidance does not imply CraftKit should auto-activate goals.
• The section stays concise and does not turn the reference into a full new skill.

Source patterns

• OpenAI Codex /goal guidance: durable objective, verifiable stopping condition, pause/resume/clear lifecycle.
• Claude Code /goal guidance: evaluator sees transcript evidence and does not call tools.
• goal-setter: intent interview plus reviewable goals/<goal-name>.md activation artifact.

[coderabbitai]

🔗 Related PRs

#3 - Add skill-radar curation layer + craft-skill-spec meta-skill [merged]
#5 - craft-handoff: pair per-session prompt with per-project rich doc [merged]
#7 - docs(craft-prompt): add /goal completion-condition reference [merged]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!