Edges

Create up to 50 edges in one atomic call. Use this for 3+ edges instead of looping `create_edge`. Edge type auto-infers when omitted.

Delete up to 50 edges in one atomic call (all succeed or all fail).

Apply up to 50 atomic re-parents. All moves validate against the schema first; any failure rolls back the whole batch.

Create one edge between two nodes. Edge type auto-infers when omitted. Target accepts an ID, or a title+type pair the server resolves. For 3+ edges, use `batch_create_edges`.

Remove one edge by ID.

Flat edge enumeration. Returns every edge of the listed `types` (or all edges when `types` is omitted) as `{id, source, target, type}` with no parent-node payload. Right for migration and canonicalisation passes. Paginates via offset/limit.

Atomic re-parent. Removes any existing hierarchy edge and creates a new one to `new_parent_id`. Validates against `UPG_EDGE_CATALOG` first; rolls back fully on failure.

Exact-match rename of every edge of type `from` to type `to`, optionally flipping source/target. Single transactional pass. Defaults to `dry_run: true`; pass `dry_run: false` to commit. Low-level primitive: skips catalog validation. Use catalog-aware migration tools for validated renames.

Inspect or drop edges whose source or target node fails to resolve. Each is classified `expected` (cross-product, sibling not loaded; keep), `suspect` (cross-product, missing product-id annotation), or `corrupt` (broken endpoint on a non-cross edge). Defaults to `dry_run: true`. Pass `dry_run: false` plus `drop: ["suspect", "corrupt"]` to remove.