llm-trace-labeling-ui

Builds a single offline HTML page that a person uses to step through a batch of LLM conversation traces, judge each whole trace as pass, fail, or defer with a free-text reason, and write every verdict to local storage the moment it is made, then verifies the page with browser screenshots and scripted checks. Use when the user asks for a hand-labeling, judging, or review tool for model outputs or agent transcripts, or wants an existing one improved. Not for multi-annotator platforms, for statistically analyzing labels already gathered, or for anything that must talk to a server.

Version

v1.0.1

Size

14 KB

Updated

2026-10-10

Your client verifies the signature before reading. This explicit manual read is available on every plan. Free automatically loads five chosen skills; paid plans automatically load relevant skills across the library.