What is idempotency in an artificial intelligence agent workflow?
Definition
Idempotency protects a workflow when a timeout leaves the caller uncertain whether an action completed. The agent or runtime can retry with the same operation key, while the receiving system recognizes the duplicate and returns the original result instead of applying the side effect again.
This matters for actions such as issuing refunds, creating records or sending commands. Idempotency requires stable request identities and durable outcome records; simply asking a model not to repeat an action does not provide the guarantee.
Acronyms and aliases
idempotent operation variant
Related terms
Frequently asked questions
Why is idempotency important for artificial intelligence agents?
Agents may retry after network errors or uncertain responses, and idempotency prevents those retries from duplicating side effects.
How is an idempotent tool call implemented?
The caller supplies a stable operation key and the service durably records and reuses the result for repeated requests with that key.
