Skip to content
XLAB / EXPERIMENT INFRASTRUCTUREWORKFLOWS

xlab-insights(7) ​

NAME ​

xlab-insights — read a report, with an optional synthesis of its evidence.

SYNOPSIS ​

sh
xlab report
xlab insight --evidence

DESCRIPTION ​

Reports are local Markdown, JSON, and self-contained HTML: runs, counts, void reasons, analyses, tree decisions, standing history, provenance, and the latest insight. The HTML includes charts and can be opened offline.

AI synthesis is optional. With an AI provider configured, the model receives numbers and labels, rather than paths, commands, keys, or system-under-test output. It must cite experiments and runs present in the evidence. Invented citations are rejected. Model-written summaries are labelled and are never measurements. Check their claims against the tables they cite.

The agent-specific evidence contract, including transcript exclusion tests, cost fronts, and a void taxonomy, belongs to planned AG-13.

EXAMPLES ​

Inspect xlab insight --evidence before enabling periodic calls. See the root README for provider configuration and key-file shape. Leave [ai] provider as none for an entirely local lab.

SEE ALSO ​

Record · Running unattended · Analysis

GENERATED FROM THE CHECKOUTSource fingerprints ↗