Negative / anti-hallucination cache (CFLAG): how it works and when to use it
Negative / anti-hallucination cache (CFLAG), records known-bad answers so every paraphrase of the question that would reproduce a hallucination is caught. Here's how Crowkis does it and why it matters for cost and safety.
The cheapest token is the one you never spend twice. Negative / anti-hallucination cache (CFLAG) is how Crowkis records known-bad answers so every paraphrase of the question that would reproduce a hallucination is caught.
How it works
Crowkis records known-bad answers so every paraphrase of the question that would reproduce a hallucination is caught. It runs inside one Redis-compatible engine, so it composes with semantic caching, agent memory, and the other intelligence layers instead of being a separate service you wire together.
CFLAG "is X free?" "yes forever" REASON "hallucinated"
Why it matters
Repetitive LLM workloads are where the money is, and semantic caching can cut costs up to 60-70% on repetitive workloads. Negative / anti-hallucination cache (CFLAG) is part of what makes that reuse safe rather than reckless, the difference between a cache you trust in production and one you audit after every incident. Drop it in over RESP, gRPC, REST, or MCP, no rewrite required.
Infrastructure earns the critical path one boring, verifiable feature at a time.