Document Browser local LLM setup

browser/index.md

@@ -102,7 +102,7 @@ FalkorDB Canvas is the standalone web component that powers the graph visualizat
 
 | Section | Description |
 | :--- | :--- |
-| Browser settings | Query timeouts, result limits, content persistence (auto-save), and display-text priority for node captions. |
+| Browser settings | Query timeouts, result limits, content persistence (auto-save), Chat LLM connection setup, and display-text priority for node captions. |
 | DB configurations (Admin) | View and update server configuration values. |
 | Users (Admin) | List users, adjust roles, add or delete users. |
 | Personal Access Tokens | Generate tokens with optional expiration and revocation management. |

browser/ui/chat-panel.md

@@ -1,6 +1,6 @@
 ---
 title: "Chat Panel"
-description: "Natural-language to query workflow, prerequisites (API key/model), and persistence limits."
+description: "Natural-language to query workflow, LLM connection setup, and persistence limits."
 parent: "UI Elements"
 grand_parent: "Browser"
 nav_order: 14
@@ -10,12 +10,24 @@ nav_order: 14
 The Chat panel lets you use English (natural language) to query the graph.
 
 ## Prerequisites
-Chat requires configuring:
-- An LLM provider API key
-- A model
+Chat requires configuring an LLM connection and a model.
 
 These are set in **Settings → Browser Settings → Chat**.
 
+You can choose one of two connection types:
+
+| Connection type | Use this when | Setup |
+| :--- | :--- | :--- |
+| **Cloud/API key** | You want to use a hosted LLM provider. | Add a provider API key, select the saved key, and choose a model. |
+| **Local LLM** | You want to use a model running locally. | Start Ollama or LM Studio, select the local provider, confirm the endpoint, and choose a model. |
+
+The LLM connection indicator in settings shows what Chat will use. It changes as you switch between cloud and local options, change the selected key/provider, or select a different model.
+
+For local models:
+- **Ollama** defaults to `http://localhost:11434`.
+- **LM Studio** defaults to `http://localhost:1234/v1`.
+- If no models are listed, check that the local server is running and the endpoint is correct.
+
 ## Opening the panel
 On the Graphs page (`/graph`), click **CHAT** in the left sidebar.
 
@@ -27,12 +39,13 @@ The number of saved interactions is configurable in settings (the UI enforces a
 {% include faq_accordion.html
   title="Frequently Asked Questions"
   q1="What do I need to configure before using Chat?"
-  a1="You need to set an **LLM provider API key** and select a **model** in Settings → Browser Settings → Chat before the Chat panel will work."
+  a1="You need to configure an **LLM connection** and select a **model** in Settings → Browser Settings → Chat. You can use either a hosted provider with a saved API key, or a local LLM through Ollama or LM Studio."
   q2="How do I open the Chat panel?"
   a2="On the Graphs page (`/graph`), click the **CHAT** toggle in the left sidebar. Note that opening Chat clears any element selection."
   q3="Can I control how many chat messages are saved?"
   a3="Yes. The number of saved interactions is configurable in **Settings → Browser Settings → Chat**. The UI enforces a bounded range."
   q4="Does Chat execute queries on the graph?"
   a4="Yes. Chat translates your **natural language** input into Cypher queries and executes them against the currently selected graph."
+  q5="Can I use Chat without a cloud API key?"
+  a5="Yes, if you use **Local LLM** with a running Ollama or LM Studio server and select one of its available models."
 %}
-

browser/ui/navigation.md

@@ -17,7 +17,7 @@ Located in the sidebar under the user name/version block:
 ## Contextual graph controls (Graphs page)
 When you are on **Graphs** (`/graph`) and a graph is selected:
 - **Graph Info toggle** (database icon) opens/closes the Graph Info panel.
-- **CHAT** toggle opens/closes the Chat side panel (requires API key/model configured in Settings).
+- **CHAT** toggle opens/closes the Chat side panel (requires an LLM connection and model configured in Settings).
 
 ## Create graph
 For non–Read-Only users, the sidebar can expose **Create Graph** (depending on the current route).
@@ -48,6 +48,5 @@ The logout button signs out and redirects to `/login`.
   q3="What does the Offline indicator mean?"
   a3="The **Offline** badge appears in the sidebar when the Browser cannot reach the backend server. Check your network connection and ensure the FalkorDB server is running."
   q4="How do I open the Chat panel?"
-  a4="On the Graphs page, click the **CHAT** toggle in the left sidebar. Note: Chat requires an API key and model to be configured in Settings first."
+  a4="On the Graphs page, click the **CHAT** toggle in the left sidebar. Note: Chat requires an LLM connection and model to be configured in Settings first. You can use either a cloud API key or a local Ollama/LM Studio connection."
 %}
-

browser/ui/settings.md

@@ -31,10 +31,37 @@ The Browser Settings panel is a page that can be scrolled, with grouped sections
 - **Graph Info**
   - Refresh interval (how often graph info refreshes)
 - **Chat**
-  - LLM provider API key
+  - LLM connection source: cloud/API key or local LLM
+  - Saved cloud API keys
+  - Local provider: Ollama or LM Studio
+  - Local endpoint URL
   - Model selection
+  - Current LLM connection indicator
   - How many interactions to store locally
 
+## Chat and LLM connection
+Use **Settings → Browser Settings → Chat** to choose what powers the Chat panel.
+
+You can use either:
+
+| Option | What it means | What you need |
+| :--- | :--- | :--- |
+| **Cloud/API key** | Chat uses a hosted LLM provider such as OpenAI, Anthropic, Gemini, Groq, Cohere, xAI, or DeepSeek. | Add a provider API key, select the saved key, then select a model. |
+| **Local LLM** | Chat uses a model running on your own machine. | Start Ollama or LM Studio locally, choose the provider, confirm the endpoint, then select a model. |
+
+The **LLM connection** area shows what Chat is currently configured to use. The status badge shows:
+- **Saved** when the displayed source, key/provider, endpoint, and model are already saved.
+- **Pending** when you changed something and still need to click **Save Settings**.
+
+For local LLMs, the default endpoints are:
+
+| Provider | Default endpoint |
+| :--- | :--- |
+| **Ollama** | `http://localhost:11434` |
+| **LM Studio** | `http://localhost:1234/v1` |
+
+If models do not appear, make sure the local server is running and the endpoint is correct.
+
 ### Replay Tutorial
 Browser Settings includes a **Replay Tutorial** action, which re-runs the guided tour overlay.
 
@@ -59,7 +86,7 @@ Allows:
 {% include faq_accordion.html
   title="Frequently Asked Questions"
   q1="How do I configure the Chat/LLM feature?"
-  a1="Go to **Settings → Browser Settings → Chat** and enter your LLM provider API key, select a model, and configure how many interactions to store locally."
+  a1="Go to **Settings → Browser Settings → Chat**. Choose **Cloud/API key** to use a hosted provider key, or **Local LLM** to use Ollama or LM Studio. Then select the key/provider, endpoint if local, and model, and click **Save Settings**."
   q2="What can Admin users do that regular users cannot?"
   a2="Admin users can access **DB Configurations** (view/update server config) and **Users** management (add/remove users, change roles). These sections require an online connection."
   q3="How do I create a Personal Access Token?"
@@ -69,4 +96,3 @@ Allows:
   q5="How do I replay the onboarding tutorial?"
   a5="Go to **Settings → Browser Settings** and click the **Replay Tutorial** action to re-run the guided tour overlay."
 %}
-