# No-op

Band: emerging — Real practice, thin evidence. Mostly vendor-stated or reported rather than measured.
Group: Documents an agent reads
Implemented by: https://clickai.dev/skills/writing-for-agents

An instruction the model already obeys by default. It pays load to say nothing. The test — does this change behaviour versus the default? — is model-relative rather than reader-relative, so two people arguing about a no-op are arguing about the default and settle it by running the document, not by debating it. When a sentence fails the test, delete the sentence rather than trimming words out of it. Related and often confused: predictable means the agent takes the same process every run, not that it produces the same output.

## See also

- [Context pointer](https://clickai.dev/legend/context-pointer)
- [Completion criterion](https://clickai.dev/legend/completion-criterion)
- [Context load / cognitive load](https://clickai.dev/legend/context-load)

---

Full legend: https://clickai.dev/legend.md