obsidian-memex-chat/README.md

Memex Chat — Obsidian Plugin

Chat with your Obsidian vault using Claude AI. Ask questions about your notes, get context-aware answers, and explore semantic connections — all without leaving Obsidian.

Features

• Vault search — TF-IDF index by default; enable local embeddings for hybrid mode (TF-IDF + semantic merged via RRF), fully offline after first model download
• Related notes sidebar — panel showing the most similar notes to whatever you have open, ranked by semantic similarity + frontmatter links + shared tags
• Auto context — relevant notes are automatically found and sent to Claude as context
• Context preview — see which notes are included before sending, or dismiss to send without context
• @mention autocomplete — pin specific notes into context directly from the input field
• Thread history — chats saved as Markdown in your vault (default: Calendar/Chat/)
• Source links — every answer shows which notes were used as context
• Prompt buttons — header mode buttons that extend Claude's system prompt (e.g. draft check, monthly review)

Installation

1. Download main.js, manifest.json, styles.css from the latest release
2. Copy into .obsidian/plugins/memex-chat/ in your vault
3. Enable in Settings → Community Plugins → Memex Chat
4. Add your Anthropic API key in plugin settings

Build from Source

npm install
npm run build

Requires Node 18+.

Usage

Basic workflow

1. Open the chat panel via the ribbon icon or the Memex Chat öffnen command
2. Type your question and press Cmd+Enter (or Enter if configured)
3. If Auto retrieve context is on, relevant notes are found automatically and shown in a preview
4. Confirm or dismiss the context, then your message is sent to Claude with the note content injected

@mentions

Type @ followed by at least 2 characters to trigger autocomplete for note names. Selecting a note adds it to the explicit context for that message, regardless of search results.

Context preview

When Show context preview is on, a list of notes appears above the input before each send. You can dismiss it to send without context, or confirm to proceed. The notes used are shown as source links below the assistant's reply.

Thread management

• Threads are listed in the sidebar of the chat panel under Verlauf
• Double-click a thread title to rename it inline
• Each thread is saved as a Markdown file in your configured threads folder, with a frontmatter id: field used for deduplication on re-import
• Vault chat files not already in active threads are loaded on demand from the sidebar

Message actions

Hover over a finished assistant message to reveal two actions:

• Copy — copies the message text to the clipboard
• Als Notiz — saves the message as a new note in Obsidian's default new-note location

Prompt buttons

Header buttons that activate a mode by extending Claude's system prompt with the contents of a vault note. Multiple buttons can be active at once.

When a button is active:

• The file at its configured vault path is appended to the system prompt
• An optional hint is shown above the input
• If searchMode: "date" is set, context retrieval switches to date-based file lookup (useful for monthly review modes)
• Auto context retrieval is skipped

Configure prompt buttons in Settings → Prompt Buttons.

System context file

In settings you can specify a vault note to always append to the system prompt (after the base prompt, before any active prompt buttons). Useful for personal context like your name, current projects, or standing instructions.

Commands

Command	Description
Memex Chat öffnen	Open the chat panel
Verwandte Notizen	Open the related notes sidebar
Memex Chat: Index neu aufbauen	Rebuild the search index
Memex Chat: Aktive Notiz als Kontext	Ask Claude about the currently open note

Related Notes Sidebar

Requires embeddings to be enabled. Opens in the right sidebar and automatically shows the top 10 most similar notes to the currently active file. Similarity is computed from:

1. Semantic embedding similarity (cosine similarity on 384-dim vectors)
2. +0.15 boost for notes linked via contextProperties frontmatter fields (e.g. related: [[Note]])
3. +0.05 per shared tag (up to +0.15)

Notes boosted by a frontmatter link are marked with a verknüpft badge.

Settings

General

Setting	Default	Description
API Key	—	Your Anthropic API key
Model	claude-opus-4-6	Which Claude model to use. Click Aktualisieren to fetch the live model list from the Anthropic API.
Max tokens	8192	Maximum output tokens per response
Max context notes	6	How many notes to retrieve per query
Max chars per note	2500	How much of each note to include
System prompt	(German default)	Base instructions sent to Claude on every request
System context file	—	Optional vault note appended to system prompt
Auto retrieve context	on	Automatically find relevant notes on send
Context preview	on	Show context before sending
Save threads to vault	on	Persist chats as Markdown files
Threads folder	Calendar/Chat	Where to save thread files
Send on Enter	off	Enter sends (vs. Cmd+Enter)
Context properties	collection, related, up, tags	Frontmatter properties whose wikilink values boost search ranking

Embeddings (optional)

Setting	Default	Description
Use embeddings	off	Enable hybrid search (TF-IDF + semantic, merged via RRF)
Embedding model	BGE Micro v2	ONNX model for local inference
Exclude folders	—	Vault folders skipped during embedding

Model	Notes
TaylorAI/bge-micro-v2	Default — fastest, 384-dim
Xenova/all-MiniLM-L6-v2	384-dim
Xenova/multilingual-e5-small	German + English
Xenova/paraphrase-multilingual-MiniLM-L12-v2	German + English, larger

Embeddings are computed locally (no API call) and cached in <vault>/.memex-chat/embeddings/. The model (~22 MB) is downloaded once to <vault>/.memex-chat/models/. Indexing progress is shown as an Obsidian notice. Obsidian Sync activity is detected automatically — indexing waits until sync is idle before starting.

Once indexing completes, context retrieval switches to hybrid mode: TF-IDF and semantic results are fetched independently then rank-merged via Reciprocal Rank Fusion. Notes that score well in both engines rise to the top; notes found by only one are still included if their rank is strong enough. This catches paraphrased queries that TF-IDF misses and avoids the over-broadness of embeddings alone.

License

MIT