Tutorial: galah (not graded — LLM required)

Upstream: 0x4D31/galah

Galah cannot be graded in this lab without --provider, --model, and API credentials (or Ollama). Do not run without usage limits on cloud LLM accounts.

When re-attempting:

git clone --depth 1 https://github.com/0x4D31/galah.git .local/labs/galah
cd .local/labs/galah && go build -o bin/galah ./cmd/galah
export LLM_PROVIDER=openai LLM_MODEL=gpt-4o-mini LLM_API_KEY=...
./bin/galah -p openai -m gpt-4o-mini -i 127.0.0.1:17081

Then add inventory + web_api_http TPS and run uhbs-lab as for HellPot HTTP labs.

No UHQS numbers are published until a full artifact tree exists.

What this skip means for analysts

This tutorial cannot reproduce a UHBS UHQS grade for this product in the current lab environment. The product hub explains the blocker. Do not invent scores. When the blocker is cleared (API keys, backend dependency, or buildable image), re-queue with the same proof pattern as graded labs: SCORECARD.txt, report.json, protocol hub with verbatim modules, and CTI/blue-team reading via READING-UHQS.md.

Trust limits

UHBS 4.2.2 evaluation notes are informative only — not certifications or endorsements. Product names appear only under conformance as survey/evaluation evidence.