Clear instructions help your Agent behave as intended. Use these basics to guide its behavior.
Tips for effective instructions
Be specific and clear
State the Agent’s role, purpose, and tasks. Avoid vague language. The more precise you are, the easier it is for the Agent to follow.
Provide examples
If you expect a format or style, include a short example in your instructions.
💡 For a summarizing Agent, show a brief model summary. For structured outputs (for example JSON), include a minimal sample object.
Test and refine
Use the preview panel to try your instructions. If the results aren’t right, edit and test again.
🔑 Iterate in small steps: change one thing, test, and repeat until the output matches your goal.
📌 Also consider the Agent’s settings:
Tone and personality: adjust the fields to guide voice and style.
Guardrails: set limits to keep responses on topic and prevent unwanted behavior.
Further reading
Good prompting improves results and consistency over time.
🔎 Grasp the best practices by reading our introduction to prompt engineering.