Markdown for LLMs

Arxo experiment 0.1

The source Markdown for this article. Copy it into your assistant or download it as a text file.

# Arxo experiment 0.1

Pinned task bank, system variants, evaluator, and runs; summary per (task, arm) with error classes, not a single score. The knowledge axis is the same agent on a new canon; the application axis is the same canon with a new agent, skill, or adapter.

## Versions

No accepted version marker is declared.

## Top-level fields

| name | type-or-$ref | required | description |
|---|---|---|---|
| `schema` | `"law.experiment/0.1"` | yes | — |
| `experimentId` | `string` | yes | — |
| `name` | `string` | yes | — |
| `axis` | enum (2) | yes | — |
| `bank` | `object` | yes | — |
| `variants` | `object` | yes | — |
| `scorer` | `object` | yes | — |
| `scope` | `object` | yes | — |
| `runs` | `array` | yes | — |
| `summary` | `array` | yes | — |

## Enumerations

| location | values |
|---|---|
| `properties/schema` | `"law.experiment/0.1"` |
| `properties/axis` | `"knowledge"`, `"application"` |

## Raw schema

[`https://law.arxo.io/schema/experiment.schema.json`](https://law.arxo.io/schema/experiment.schema.json)