With Relations
PAM supports semantic relations between memories via the relations array. Each relation specifies a directional link
between two memories with a type and optional confidence score.
Example
Section titled “Example”{ "schema": "portable-ai-memory", "schema_version": "1.0", "export_date": "2026-02-17T00:00:00Z", "owner": { "id": "user-123" }, "memories": [ { "id": "mem-001", "type": "skill", "content": "User is a cloud infrastructure engineer", "content_hash": "sha256:e1bae3ec291c99eced01fc91b4152a0cef541fccf2034fc11b3f90f4e4d79b6e", "temporal": { "created_at": "2026-02-15T00:00:00Z" }, "provenance": { "platform": "chatgpt" } }, { "id": "mem-002", "type": "preference", "content": "User prefers terraform over cloudformation", "content_hash": "sha256:85613bb561dda27606c3a2f3eb0e606f40f53af0fab0da60b2b95ed92ac1d890", "temporal": { "created_at": "2026-02-15T00:00:00Z" }, "provenance": { "platform": "chatgpt" } } ], "relations": [ { "id": "rel-001", "from": "mem-002", "to": "mem-001", "type": "derived_from", "confidence": 0.9, "created_at": "2026-02-15T00:00:00Z" } ]}Relation Types
Section titled “Relation Types”| Type | Semantics |
|---|---|
supports | Source provides evidence for target |
contradicts | Source conflicts with target (useful for tracking belief changes) |
extends | Source adds detail to target |
supersedes | Source replaces target |
related_to | General semantic relationship |
derived_from | Source was inferred from target |
Fields
Section titled “Fields”| Field | Type | Required | Description |
|---|---|---|---|
id | string | yes | Unique identifier for the relation |
from | string | yes | Source memory ID |
to | string | yes | Target memory ID |
type | string | yes | Relation type (see above) |
confidence | number | null | no | Confidence in the relationship [0.0–1.0] |
created_at | string | yes | ISO 8601 timestamp |
See spec §13 for full relation semantics.