Using Grimoire
Plain overview of the features that ship in 1.x. Nothing here is aspirational.
Save links
Add a URL from the library (or paste into the main page). Grimoire rejects private/loopback hosts and non-http(s) schemes. Duplicates of an active bookmark return the existing item. Links already in archive or trash need restore or permanent delete first.
New bookmarks show up immediately as saved, then a background job fetches and processes them. Failures leave the bookmark usable; you can retry from the detail view or reprocess batches in Settings.
Content extraction
After save, Grimoire pulls readable content when it can:
| Source | What it stores |
|---|---|
| Normal web pages | Cleaned text / Markdown |
| PDFs | Extracted text |
| GitHub repositories | Metadata + README |
| GitHub issues | Issue body, labels, top comments |
| Stack Overflow / Stack Exchange | Question and top or accepted answer |
| YouTube | Metadata and captions when available |
Organize
- Categories — create, rename, nest (max three levels), drag to reorder or reparent
- Tags — create, rename, attach; browse from Tags pages
- Per bookmark — title, notes, pin, read/unread, read later, archive, trash, restore, permanent delete
- Bulk — select bookmarks to delete or move
- Filters — category, tag, domain, date, and related library filters
Optional AI can suggest summary, tags, and a broad category during ingest. When an embedding provider is configured, an organization agent may propose duplicates or similar categories; review them in Review Queue.
Search
| Mode | When it works |
|---|---|
| Keyword | Always (SQLite FTS5 over title, summary, tags, content) |
| Semantic | Embedding provider configured |
| Hybrid | Embedding provider configured (keyword + vector + recency) |
Related bookmarks use embeddings when available.
Capture from the browser
There is no Chrome/Firefox store extension in 1.x yet. Use the built-in bookmarklet:
- Open Settings → Browser Integration
- Create an integration token (the full secret is shown once — copy it if you need it elsewhere)
- While that new token is still on screen, use copy the bookmarklet URL (or the Bookmarklet button on that token row)
- In your browser, create a bookmark and paste the copied
javascript:…URL into the bookmark’s URL field
Existing tokens only store a prefix. You cannot regenerate a bookmarklet for an old token — create a new one. The bookmarklet embeds the token; treat it like a password.
You can also import a Netscape/HTML bookmark export from your browser, and export your library as JSON or CSV.
Backups
From Settings or the littleimp CLI you can create, list, verify, and restore local snapshots. Encrypted packages and optional scheduled/S3 targets are available in Settings when configured. Restores verify checksums and create a rollback copy first.
Local integrations
- REST — see API.md; health at
GET /health - MCP — Streamable HTTP at
http://127.0.0.1:3210/mcpwith an integration bearer token - Capture API — token-protected
POST /capturefor same-machine clients
Privacy defaults
Data stays under your local data directory (Docker volume, or ~/.local/share/littleimp/ for native installs). The daemon binds to loopback by default. External AI is opt-in; without it, nothing leaves your machine for enrichment or embeddings.
More answers: FAQ.