Browse the documentation

REST API

Compute and snapshot

Read the computed values of a Layerz model over REST, and download or upload the whole model as one JSON document with optimistic locking.

Compute

GET /models/:id/compute computes the model and returns every line's series. ?items=uid1,uid2 limits the response to some lines.

{
  "values": { "a1b2c3d4": [100, 120, 140] },
  "periods": { "yearly": ["2027", "2028", "2029"], "constant": ["constant"] },
  "item_timelines": { "a1b2c3d4": "yearly" },
  "order": ["e5f6a7b8", "a1b2c3d4"],
  "errors": {}
}
  • values[uid] aligns with periods[item_timelines[uid]].
  • A line computed once (a constant, an IRR, an NPV) has one value, on the constant timeline.
  • When the model cannot compute (a cycle, a broken reference), the response adds computable: false and compute_errors (code, message, item_uids), and the affected series read zero. Fix the model before reading its numbers.

Values are the base plan. Scenario views, values at another grain and the check report are available over MCP (layerz_read with branch_id, granularity, health).

Snapshot

The snapshot is the whole model in its stored format: { manifest, schema, inputs }.

  • GET /models/:id/snapshot returns it with an ETag (a hash of its content).
  • PUT /models/:id/snapshot replaces the model with the document you send. Headers:
Header
X-Version-LabelRequired: the label of the version this write records
If-MatchOptional: the ETag you read; the write fails with 409 if the model changed since

The model is validated and repaired at write time like any other write. Use the snapshot to keep a model in a repository, diff it in code review, or move it between accounts.