JSON output

Learn how AgentIdem Nebutex structured output can be used in scripts, CI systems, and other automated tooling.

JSON output

AgentIdem Nebutex uses structured data for traces and reliability results so they can be consumed by automation and external tooling.

JSON is useful when AgentIdem output needs to be:

  • stored as a CI artifact
  • inspected programmatically
  • processed by scripts
  • passed to another tool
  • compared later
  • archived for debugging

Structured test results

AgentIdem reliability testing produces structured report data.

A report can include information such as:

  • target
  • baseline result
  • fault results
  • findings
  • invariant results
  • unsafe count
  • safety status

This keeps machine-readable results separate from human-readable CLI presentation.

Structured traces

Recorded traces also use structured data.

A trace can contain operation information such as:

  • operation name
  • operation kind
  • arguments
  • logical identity
  • result
  • error
  • execution status
  • observation state
  • timestamps

For example, a write operation may conceptually contain:

{
  "name": "refund",
  "kind": "WRITE",
  "identity": "payment-123",
  "status": "SUCCESS",
  "observation": "RECEIVED"
}

The exact serialized structure depends on the AgentIdem Nebutex trace schema.

Recording traces

Use agentidem record to execute a target and save its trace.

agentidem record examples.refund_agent:run --trace trace.json

The resulting trace can then be inspected or used with replay.

Replay input

A saved trace can be passed to agentidem replay.

agentidem replay examples.refund_agent:run --trace trace.json

Replay compares meaningful operation behavior against the recorded trace.

Stable vs volatile fields

Some trace values describe meaningful execution behavior.

Examples include:

  • operation name
  • kind
  • arguments
  • identity
  • result
  • error
  • status
  • observation

Other values may be volatile between executions.

Examples include:

  • trace IDs
  • operation IDs
  • timestamps

Replay should focus on meaningful operation behavior rather than requiring volatile metadata to remain identical.

CI and automation

Structured AgentIdem output can be combined with CLI exit codes.

For example:

agentidem test examples.refund_agent:run

The exit code indicates whether the command succeeded, found an unsafe result, or encountered an operational error.

Structured output can provide the detailed evidence behind that result.

See Exit codes.

Schema compatibility

When consuming AgentIdem JSON from another program, avoid depending on undocumented internal fields.

Use documented structured fields and schema versions where available.

This makes integrations more resilient as AgentIdem Nebutex evolves.

Next steps