Grafting moves useful memory between sessions.

Grafting assembles prompt-ready context from selected topic nodes or copies active memory into another session with provenance.

When grafting helps

Grafting is for moving relevant memory into a new working context without copying an entire transcript. It is useful when a user changes sessions, a worker needs shared context, or an assistant needs only the parts of a graph that match a task.

Every grafted path should remain explainable. MemoGrafter keeps provenance so Studio and debugging tools can show where copied memory came from.

Two paths

Explicit grafting selects known topic IDs and formats them into a prompt.

Semantic grafting finds topic seeds by query similarity before assembling the prompt.

Absorption copies active topic nodes and active memory nodes into a target session.

Graft flow

Source memory

Select topics

Provenance

Target session