How to Design Cache Keys for LLM API Responses
A prompt-only cache key can return stale data or expose a response across application scopes.
Include result-changing inputs
A cache key may include:
tenant or visibility scope;
task identifier;
mo
vectronode.hashnode.dev1 min read