📖 My Glossary — User Guide

1. Getting started

My Glossary is one personal glossary with several doors into it:

1. Send /start to the bot — it replies with your Telegram ID.
2. Paste that ID into the extension settings (＋ Add panel) — your glossary syncs automatically.
3. Optional but recommended: add your Claude API key in Web App Settings (⚙️) and in the extension — it unlocks every AI feature. The key is stored only on your device and never sent to our server.

2. Adding terms

From a web page (extension)

Select text on any page → right-click → “Add to Glossary”. The popup opens with the term pre-filled. Click ✨ to generate a definition (in the language picked in the extension settings) — AI also suggests which groups the term belongs to.

If no existing group fits (or you have no groups yet), AI proposes a brand-new one — it appears as a dashed amber chip ➕ name. Keep it and the group is created the moment you save the term; tap × to dismiss. The same works in the Web App's ✨ Generate. Suggestions never override groups you picked yourself.

Two more helpers in the extension form: the ✨ button next to the group select suggests groups on demand (instant name match first, then AI — works even after you've already picked some), and the “↑ Parent node” search field attaches the new term under any existing term — start typing, pick a suggestion (a unique partial match works too), and it lands in the right mind-map branch right away.

From any app (Android)

In the Android app you can add a term from anywhere on the phone: select text in any app (browser, reader, chat) → Share → My Glossary. The Add screen opens with the term name pre-filled — tap ✨ for a definition and save. It's the phone-wide equivalent of the Chrome extension.

In the Web App

＋ button on the home screen: name, definition (✨ generates it), groups as tap-to-toggle chips, tags. After saving, AI quietly enriches the term in the background: difficulty score, a structure if the term is list-like, a parent in the knowledge tree, and semantic relations to your existing terms.

Duplicate protection

While typing in the extension you'll see a live hint if a similar term already exists (exact or fuzzy match). The Web App asks for confirmation before saving a near-duplicate.

Browsing: tiles and the term screen

Terms are shown as tiles — a responsive grid on desktop, a single column on the phone. A tile shows only the term name; tap it to flip to the definition (a quick self-check without entering review mode). The ↗ button in the tile corner opens the term screen — the full card with everything: definition, structure, AI notes, groups, tags, related terms, breadcrumbs, actions and chat. Breadcrumbs, children and siblings on the term screen are links — tap to jump between terms like in a wiki; ← Back returns where you came from.

3. Spaces, groups, subgroups & priorities

Spaces — shelves above groups

When groups pile up, organize them into spaces (📚) — top-level shelves like “Networking”, “English”, “Work”. The space row sits at the top of the home screen: All · 📚 your spaces · 📦 Unsorted · ➕ space. The active space filters everything — visible groups, term tiles, the Review button and the mind map overview — and is remembered between visits. A root group is assigned to a space on its group screen (📚 Space select) or simply by being created while that space is active; subgroups inherit the parent's space. 📦 Unsorted shows groups without a shelf and terms without groups. Deleting a space (🗑 when it's active) keeps all its groups and terms — they just leave the shelf.

Groups

• Create: “➕ group” on the home screen, in the bot (/groups), or in the extension's group select.
• Rename: open a group → ✏️ next to its name.
• Subgroups: open a group → “➕ subgroup”. Two levels max. Deleting a parent makes its children top-level; terms are never deleted with a group.
• Re-parent: the “↳ Parent group” select on the group screen turns a top-level group into a subgroup of another (or back to top level with “⌂ Top level”). Only root groups can be parents (2-level cap), a group that already has its own subgroups can't be nested, and a group loses its own shelf when it becomes a subgroup (it inherits the parent's space).
• A term can be in several groups. Edit them on the term screen (↗ on a tile): × on a chip removes, “＋ group…” adds, and ✨ Suggest groups picks them automatically — instant name match first, then AI (which may create one new group).
• Priorities: in the extension, open the 📌 panel — ▲▼ arrows rank groups. The rank (weight 1.0 → 0.5) decides which group wins in autosuggestion and orders the group chips of every term. Synced to the server, shared by all devices.
• Pins: ★ in the same panel pins a group to the extension's filter bar.
• Review a single group: “🔁 Review this group” on the group screen, or the selector inside review mode.

Structure map (🗂)

The 🗂 button in the home header opens a Miro-style overview of how everything is organized: 📚 spaces on the left, their 📁 groups next, ↳ subgroups after that, with the term count on each node. It's interactive — drag a group onto a 📚 space to make it a top-level group there, or onto another 📁 group to turn it into that group's subgroup. A group that already has its own subgroups can't be nested (2-level limit). Great for seeing the big picture and reshuffling branches at a glance.

4. Tags

Tags are light, cross-cutting labels — a term can have any number of them (groups are for hierarchy and review; tags are for filtering).

• Create / attach while adding: type in the Tags field, press Enter (or comma). Existing tags autocomplete; colors are assigned automatically.
• On an existing term: open the term screen (↗) → tag row → “＋ tag…” select (or “➕ New tag…”), × on a pill detaches.
• Filter: tap tag pills on the home screen — multiple selected tags work as AND.
• In bulk: ☑ Select mode → “Add tag…” in the bottom bar.

5. Reviewing: cards, quiz & sort

Spaced repetition (FSRS): answer honestly and cards return less and less often. A term counts as learned after 3 successful reviews in a row. Morning reminders arrive in Telegram at the hour you set.

Modes (toggle at the top of the review screen)

• 🃏 Cards — classic flip card: tap to reveal, then “✅ Know / ❌ Don't know”. The back also shows the term's mnemonic and structure if present.
• ❓ Quiz — 4-option multiple choice (term→definition or definition→term). Needs at least 4 terms; the correct answer lights up green. In the bot: /quiz with A–D buttons.
• 🔀 Sort exercise — terms with an ordered structure (like the OSI model) ask you to arrange shuffled items: drag by the ⠿ handle or use ▲▼. A perfect order counts as “know”, any mistake as “don't know”.

Difficulty affects scheduling

When AI rates a new term (1–5 dots in the card), the first review is scheduled accordingly: easy terms (●●) wait 2 days, medium (●●●) 1 day, hard (●●●●+) show up immediately.

6. AI features in the card

Open a term (↗ on its tile) to find AI helpers (all need your Claude key; results are cached on the server, so nothing regenerates on every open):

7. Structured terms

Some terms are inherently a list or a table — the OSI model, HTTP methods, device families. Such terms get a structure:

• AI proposes it automatically when you add a list-like term, or on demand via 📋 in the card (🔄 regenerates, × removes).
• Ordered lists render numbered, unordered as bullets (with optional notes), tables as real tables — in the card and on the back of review cards.
• Ordered structures unlock the sort exercise in quiz mode.

8. Mind Map

Your glossary is also a knowledge tree: every term may have a parent. The tree is built three ways: AI suggests a parent when you add a term, the “↑ Attach to a parent node…” search field on the term screen and in the extension's add form (type to filter, pick a suggestion; “⌂ Make root” detaches, cycles are rejected), and ✨ Expand on the map below. The map makes the tree visible.

• Open: 🗺 in the home header (whole forest) or “🗺 Mind map” on a term screen (centered on that term). Tapping the central node opens the term screen back.
• Centered view (Miro-style): the term is the central node, fundamentals branch left (ancestors, cyan), applied concepts branch right (descendants). Related-but-unlinked terms hang below on dashed amber edges.
• Navigate: tap any node to re-center on it; tap the central node to open its card. ⌂ Overview returns to the full view, where tapping a branch collapses it.
• Node colors: purple = a term in your glossary (the focus is bold purple, fundamentals are cyan), gray dashed ➕ = an AI suggestion that is NOT in your glossary yet.
• ✨ Expand: asks AI to grow the focused node — one more fundamental parent (if missing) and 3–5 applied children. They appear as gray ghost nodes; nothing is saved yet.
• Tap a gray node to add it: the term is created with its definition, linked into the tree, and the node turns purple. Groups are assigned automatically — inherited from the anchor card, or matched by name if the anchor has none. Re-focusing or ⌂ Overview discards the remaining ghosts.
• Pinch-to-zoom and drag work on mobile.

9. Card chat — AI with tools

💬 Chat on the term screen opens a mini-dialog with Claude Sonnet that sees the card (definition, explanation, mnemonic, structure). Ask anything: comparisons, examples, “why is it called that”.

The chat can also act:

• “Fix the definition — it confuses L2 and L3” → ✏️ the card is updated (name, definition, explanation, mnemonic, difficulty).
• “Add the main gateway models as sub-terms” → 🌱 up to 5 child nodes appear on the mind map as real cards.

Every change is reported as a green line in the chat log and the card refreshes immediately. Replies render with full markdown — lists, code blocks and comparison tables display properly. History is kept per term on your device; 🗑 clears it.

10. Bulk editing & cleanup

• Web App: ☑ Select in the home header → check terms (or Select all) → bottom bar: move to a group, add a tag, or delete (with confirmation).
• Bot: /cleanup walks through terms that have no group, 5 at a time, with one-tap “move to group” buttons.

11. Telegram bot commands

12. Sync & the Chrome extension

• The extension pushes new terms to the server as you add them (☁️ indicator in the header) and re-pushes anything pending when the popup opens.
• ⟳ Sync pulls everything from the server: terms you added in Telegram, group changes made in the Web App. On a fresh browser profile just enter your Telegram ID — a full sync runs automatically.
• The newest extension build is always at passiveminde.com/extension.zip — the zip is assembled from the current sources on every download.

13. Accounts & sign-in

• Email & password — the standalone way, used by the Android app and the web. On the login screen tap Sign up, enter an email and a password (8+ characters), and you're in.
• Already use Google? To log in on the phone with the same account, open the Web App in a browser, sign in with Google, then Settings → Set a password. After that use that email + password in the Android app — it's the same account, all your terms are there (no import needed).
• Inside Telegram no login is needed.
• In a regular browser you can also Sign in with Google at /app.
• To make a web/email account see your Telegram glossary: send /link to the bot, then enter the 6-digit code in Web App Settings → Link Telegram. After that all clients share one glossary.

14. Backup: export & import

Web App Settings → Backup saves your whole glossary to a file and loads it back — on the same account or a different one.

• Export JSON downloads glossary-export.json: every term with its definition, difficulty, structure, group/tag membership, parent (wiki-tree) hierarchy and relations. Everything is referenced by name, so the file is portable between accounts and devices.
• Import JSON picks a file and loads it. Terms you already have (matched by name, case-insensitive) are skipped, so re-importing your own export changes nothing — it's safe to run twice. Missing groups, tags and spaces are created automatically.
• After an import you get a summary: how many terms, groups, tags and relations were added or skipped, plus any warnings (for example, a relation that points at a term not present in the file).