Skip to content
Glossary / Context and meaning preservation

Retrieval mutation

Retrieval mutation is any meaningful distortion introduced between an original source and the retrieved context a system actually uses, including truncation, smoothing, reframing, selective quoting, or a dropped decisive qualifier. The retrieved text can look faithful while no longer meaning what the source meant, and the system then reasons over the mutated version as if it were the source.

How it manifests · Tool-addressed

It shows up when a retrieval or summarization step truncates or reframes a source so the retrieved text no longer means what the source meant. fidelis avoids it on the memory path by returning the original text rather than a regenerated version.

Evidence: fidelis

Related terms

All terms · Where we work