Define what the model must refuse and how it says no. Use when shipping a user-facing agent or tool-calling loop.
--- name: write-a-refusal-policy category: ai-agents description: >- Define what the model must refuse and how it says no. Use when shipping a user-facing agent or tool-calling loop. --- # Write a Refusal Policy List disallowed intents. Give one short refusal template. Never continue after a hard refuse. Log the category, not the raw user text, when privacy requires it.
Shared publicly by the author and open for any use. See the Terms.