project-nomad · Issues· 93 open

Dark mode: headings and the Builder Tag control fail WCAG contrast across the settings pages

Link tile with a URL over 255 characters returns a 500 with raw SQL in the log

The new relevance floor does not reject coherent off-topic queries, and fails entirely on general-prose corpora

[Bug]: new install will not download AI models tried discord but it will not let me post.

[Bug]: iGPU passthrough on AMD mini pc shows a warning "GPU Not Accessible to AI Assistant,"

Ollama CPU-fallback detection cannot fire on AMD systems

Chat answers are silently truncated at 1024 output tokens

fix(downloads): a single stall fails the job permanently, bypassing attempts: 10

file-embeddings stalled-job recovery forks the batch chain and silently writes duplicate vectors

Model download and embed dispatches dedupe onto retained finished jobs (same add-dedupe behavior as #1213)

Failed Ollama model pulls are reported as successful: in-band NDJSON errors are silently dropped

[Bug]: AI not installing

_embedWithFallback() ignores isOllamaNative, always tries /api/embed first even when known to be non-Ollama

fix(system): Settings > System still shows raw PCI id as the GPU model on AMD

[Feature]: Surface source document + date as citations under AI chat answers