Claude Code /goal vs Codex /goal
Both keep an agent working toward a goal with validation and a stop condition. Claude Code /goal frames a single goal with explicit boundaries and a max-iteration line; Codex /goal frames an objective with a “Done when” block and explicit stop conditions. The structure differs more than the intent.
Why it matters
Using the format your tool expects reduces ambiguity. A Claude Code /goal pasted into Codex still works, but matching each tool's conventions makes the goal, validation, and stop rule unmistakable.
Claude Code /goal vs Codex /goal
| Aspect | Claude Code /goal | Codex /goal |
|---|---|---|
| Framing | A single goal to work toward | An objective with a validation loop |
| Done condition | Validation block | “Done when” block |
| Stop rule | Stop rule + Maximum iterations | Stop conditions (checks pass / N failures / budget) |
| Boundaries | Boundaries block | Constraints block |
| Context | Context block | Context block |
| Maturity | Official (Claude Code) | Official (Codex) |
Which format to copy
- 01Working in Claude Code?Claude Code /goal format
- 02Working in Codex?Codex /goal with “Done when”
- 03Another agent or tool-agnostic?Generic Agent prompt
- 04Want a runtime to enforce caps?continuous-claude command
Practical checklist
- State an outcome, not steps, in both
- Give a machine-checkable validation or done condition
- List boundaries / constraints explicitly
- Set a hard stop: iterations or repeated failures
- Require human approval before irreversible actions
Example
In the Goal Generator, the same form produces a Claude Code /goal (with a Stop rule + Maximum iterations) and a Codex /goal (with “Done when” + Stop conditions). Copy whichever your tool uses.
Common failure modes
Related templates
FAQ
It usually still works — the intent is the same. But matching each tool's conventions (“Done when” for Codex, “Stop rule” for Claude Code) makes the goal and stop condition unmistakable.