Most MCP memory servers store text. Few tag where it came from.

Look across the MCP memory servers people actually run and almost all of them share one property: once something's saved, it's just "a memory." Nothing in storage distinguishes a fact you stated from a document it read from a summary it wrote itself. Six of the most-used servers, compared on exactly that.

Six servers, compared

Read from each project's own README and docs as of publish time; check their repos directly before relying on specifics. server-memory · mcp-memory-service · Mem0 · Letta · Zep

The distinction that gets lost

Three things can end up in a memory store, and they carry very different weight: something you told the assistant directly, something the assistant read from a document you gave it, and something the assistant itself concluded or summarized from other memories. Stored as undifferentiated text, all three come back identically confident. A model-generated summary that drifted slightly from the source looks exactly as trustworthy as the sentence you typed yourself, because nothing in the data says otherwise.

Why it matters specifically when a memory is wrong

This stops being an abstract data-modeling concern the moment a recalled memory turns out to be stale or incorrect. Which of the three types it was determines what the right fix even looks like: correcting a stated fact you'd change your mind about is different from re-reading a source document that changed, which is different again from a summary that needs to be regenerated once its inputs move. A memory store that never captured which one it was can't tell you which fix applies; it can only tell you something doesn't match anymore. What a forget call actually has to guarantee covers the failure modes that follow from exactly this gap.

How ModelBrain does it instead

Every memory ModelBrain stores carries a provenance tier at write time, source evidence, your own stated assertion, or a model-generated summary, kept as a distinct, queryable field rather than collapsed into one opaque blob. It's a small addition to the schema and a real constraint on the write path, not a retrieval-time trick layered on afterward. How ModelBrain tags what it remembers has the actual mechanics.