Frequently Asked Questions

GNO is a local knowledge engine that indexes your documents—Markdown, PDF, DOCX, XLSX, PPTX—and lets you search them using BM25 full-text, vector similarity, or hybrid search with query expansion and cross-encoder reranking. Everything runs 100% locally on your machine with zero cloud dependencies.

Install via Bun with bun install -g @gmickel/gno. On macOS, vector search requires Homebrew SQLite (brew install sqlite3). Run gno doctor to verify your installation, or launch gno serve for a guided first-run setup in the browser.

Yes. After install, run gno serve and open localhost:3000. The dashboard walks you through adding a folder, choosing a speed-vs-quality preset, downloading local models if needed, and checking health for indexing, models, and disk.

Yes. Run gno daemon for headless continuous indexing. It keeps the same watch, sync, and embed loop alive without starting the Web UI. In v0.30 it stays in the foreground, so use nohup, launchd, or systemd if you want supervision.

GNO indexes Markdown (.md) with frontmatter extraction, PDF (.pdf), Microsoft Word (.docx), Excel (.xlsx), PowerPoint (.pptx), and plain text files. It also detects document language across 30+ languages.

Not in place. GNO can view and search converted PDF/DOCX/PPTX/XLSX content, but those source files stay read-only. If you want to work on the content, use Create editable copy to generate a markdown note with source provenance.

No. GNO is 100% local and privacy-first. All indexing, embedding, search, and AI answers run on your machine. Zero telemetry, zero tracking. Your documents never leave your computer. It works completely offline.

Yes. Point GNO at your Obsidian vault directory and it indexes everything. GNO understands wiki links ([[note]]), frontmatter tags, and can discover semantic connections between notes that Obsidian’s built-in search misses. It adds hybrid search, AI answers, and a knowledge graph on top of your existing vault. The app now previews likely Obsidian vault imports before indexing so users can see what will happen first.

Yes. In the app, use Collections > Add Collection and GNO will preview the folder before indexing. It detects likely Obsidian vaults, mixed work-doc folders, and obvious duplicate conflicts so you can understand the import before it starts.

gno search uses BM25 keyword matching (instant, ~5-20ms). gno vsearch uses vector similarity for semantic meaning (~0.5s). gno query combines both with query expansion, reciprocal rank fusion, and cross-encoder reranking for best results (~2-3s default, ~0.7s fast mode).

Use gno ask “your question” –answer. This searches your documents using the hybrid pipeline and generates a cited answer using a local LLM. No cloud APIs needed.

Yes, use –collection or -c flag, for example gno search “term” -c notes. You can also filter by tags (–tags-any, –tags-all), dates (–since, –until), and exclude specific documents (–exclude).

GNO supports multi-line query documents using term: (exact keywords), intent: (semantic intent steering), and hyde: (hypothetical document passages). These give you explicit control over the retrieval pipeline for complex queries.

On macOS and Linux it is at ~/.config/gno/config.yaml. On Windows it is at %APPDATA%\gno\config.yaml. Run gno doctor to see your exact path.

Use gno collection add path –name name for each folder, then run gno update to index them all. You can add Obsidian vaults, project docs, research folders, and code repos—any directory on your machine.

Use gno serve when you want the browser dashboard, REST API, or desktop shell integration. Use gno daemon when you only want headless continuous indexing. In v0.30 avoid running both against the same index at the same time.

Yes, add them to the exclude array in your collection config, for example exclude node_modules, .git, and temporary files. You can also use the –exclude flag at query time to skip specific documents from results.

GNO uses local models via node-llama-cpp. The default preset is slim-tuned (~1GB) with the fine-tuned local expansion model plus the same embed, rerank, and answer stack as slim. Other built-in presets are slim, balanced (~2GB), and quality (~2.5GB).

No. GNO runs entirely locally with no API keys or cloud services required. Models are downloaded once and run on your machine.

Use gno models use preset then gno models pull to download, or switch presets directly in the web UI. Presets are slim-tuned, slim, balanced, or quality. You can also offload inference to a remote GPU server on your network via HTTP backends.

GNO publishes a fine-tuned retrieval model (gno-expansion-slim-retrieval-v1) on Hugging Face, trained locally on Apple Silicon with MLX LoRA. It achieves nDCG@10 of 0.925 and Ask Recall@5 of 0.875. Install it with gno models use slim-tuned.

Install GNO as a skill with gno skill install –scope user. This works with Claude Code, Codex, OpenCode, and OpenClaw. Your agent can then search your knowledge base on demand with zero context window pollution.

Install GNO as an MCP server with gno mcp install –target cursor (or zed, windsurf, amp, etc.). GNO provides 19 MCP tools for search, retrieval, indexing, and graph exploration.

Model Context Protocol (MCP) is a standard for AI assistants to access external tools. GNO provides an MCP server so Claude Desktop, Cursor, Zed, Windsurf, Amp, Raycast, and more can search your documents.

Yes, use –json flag for machine-readable output, for example gno search “term” –json piped to jq. GNO also supports CSV, Markdown, XML, and URI output formats. For TypeScript apps, use the SDK to embed GNO directly.

Yes. The SDK lets you import GNO directly into any Bun or TypeScript app with createGnoClient(). Run search, query, ask, index, and more without CLI subprocesses or a local server.