Limitations
NoDrift reduces risk. It does not promise perfect output, total detection, or automatic readiness.
What Not To Infer
NoDrift does not guarantee that every mistake will be prevented or detected. It does not control the LLM, and it does not turn protected workspace materials into outgoing prompt content.
Evidence Has Limits Too
Confirmed zeros do not establish complete detection while audit coverage is incomplete, and candidate incidents stay separate from final classifications until the evidence supports a conclusion.
Docs Do Not Replace The Workspace
Protected procedures, private records, and detailed operating language remain in the workspace. They are not part of the public documentation surface.
Gated Claims Stay Gated
Commercial terms, support-readiness language, package internals, external-state claims, and future-tool availability remain gated until they are explicitly approved and verified.