MatrixCairn is a Windows AI coding client for examining matrix code in its project context. Open the relevant folder, discuss shape rules with a compatible model and inspect the suggested loops before applying them.
def matrix_product(left, right):
left, right = [tuple(r) for r in left], [tuple(r) for r in right]
widths = []
for matrix in (left, right):
width = len(matrix[0]) if matrix else 0
if len(matrix) > 64 or width > 64 or any(len(r) != width for r in matrix):
raise ValueError("matrix.shape")
if any(type(v) is not int for r in matrix for v in r):
raise ValueError("cell.integer")
widths.append(width)
if widths[0] != len(right):
raise ValueError("matrix.incompatible")
return [[sum(row[k] * right[k][j] for k in range(widths[0]))
for j in range(widths[1])] for row in left]Compare actual behavior with the expected result. If a check fails, investigate it before applying more changes.
Base URL, credentials and a reachable server.
Test empty and single-cell inputs, rejecting ragged rows, incompatible dimensions and Boolean cells without modifying the matrices.
def matrix_product(left, right):
left, right = [tuple(r) for r in left], [tuple(r) for r in right]
widths = []
for matrix in (left, right):
width = len(matrix[0]) if matrix else 0
if len(matrix) > 64 or width > 64 or any(len(r) != width for r in matrix):
raise ValueError("matrix.shape")
if any(type(v) is not int for r in matrix for v in r):
raise ValueError("cell.integer")
widths.append(width)
if widths[0] != len(right):
raise ValueError("matrix.incompatible")
return [[sum(row[k] * right[k][j] for k in range(widths[0]))
for j in range(widths[1])] for row in left]MatrixCairn provides a Windows AI coding workspace for working on code that accepts structured numeric data. Bring the relevant project files and describe the allowed shapes, value types and result expected by the caller. Ask the model to inspect validation and computation together. This makes it easier to review whether a proposed change preserves the structure your application needs, even when individual calculations look reasonable.
For a task involving matrix.py, you might examine how incompatible dimensions or uneven rows are rejected before matrix multiplication. Ask the assistant to explain the current assumptions and propose a narrowly scoped correction if one is needed. Use explicit input and output examples, including the empty shapes your project permits. The purpose is to improve your implementation with reviewable evidence, not to infer mathematical correctness from a tidy function.
Run the project checks after inspecting the changes, and confirm how invalid input reaches the consumer. Include the boundary sizes you actually support rather than relying only on a small successful example. An unfiltered model can be connected through a supported OpenAI-compatible or Anthropic endpoint when offered, with credentials where required. MatrixCairn leaves tools, context capacity and model restrictions dependent on that configured service; the software workspace supports the discussion while you verify the calculation.
Define whether the application rejects it or transforms it before calculation. Do not let an incidental iteration rule decide. Ask the assistant to inspect where that policy belongs and verify the error observed by the caller.
Also check output dimensions and the correspondence between rows and columns. A few correct values do not establish that the result has the shape expected by downstream code, particularly after changes to empty-input handling.
State the supported bounds and inspect where they are enforced. Test accepted boundary sizes and an oversized case. Limits in an example are not a performance guarantee for your application or permission to skip runtime checks.
Decide whether that type belongs in the contract, then inspect conversion or rejection explicitly. A permissive change can affect arithmetic and serialization later, so review the full path rather than broadening accepted values without a reason.
MatrixCairn · Updated 2026-10-25