galah — skipped (LLM dependency)

Status: Informative · not graded
Upstream: https://github.com/0x4D31/galah

Galah is an LLM-powered HTTP honeypot: every novel request can invoke a configured provider (OpenAI, Google AI, Anthropic, Cohere, GCP Vertex, or Ollama) to synthesize headers and bodies. The upstream CLI requires --provider and --model, and documents LLM_API_KEY / cloud credentials. This UHBS lab session had no OpenAI, Anthropic, or other LLM API keys in the environment, and spinning up a local Ollama model was out of scope for the air-gapped Docker grading window.

Without a live model, the service cannot answer UHBS HTTP fidelity and behavioral probes in a reproducible way; scores would reflect infrastructure failure, not decoy quality. UHBS therefore records a skip note rather than fabricated UHQS values. Re-queue when operators can supply keys with billing limits, or when a pinned local Ollama image is acceptable inside an isolated lab VLAN.

See TUTORIAL.md · METHODOLOGY.md.

Trust

Informative UHBS 4.2.2 gap note — not an endorsement. READING-UHQS.md