Open Agent Loops
API Reference

formatCodeExecutionResult

@open-agent-loops/core


function formatCodeExecutionResult(result): string;

Defined in: tools/builtin/code-execution.ts:78

Fold a CodeExecutionResult into the single text result the loop hands to the model.

Parameters

ParameterTypeDescription
resultCodeExecutionResultThe stdout/stderr/exit code captured by a CodeExecutionBackend.

Returns

string

A text block: stdout, an annotated stderr section when present, and an always-present exit verdict.

Remarks

Every result ends with an explicit verdict — [exit 0: ok] on success or [exit N: error] on failure — so a code run is never a contentless result: "ran but printed nothing" reads as a clear success rather than an ambiguous empty string. A non-zero exit is NOT a tool error (the code ran); it is surfaced so the model can react. This deliberately diverges from formatShellResult, which hides a zero exit and falls back to "(no output)" — code execution always states its outcome.

See

codeExecutionTool

On this page