sync

docs/features/plugin/tools/index.mdx

@@ -137,6 +137,32 @@ These models excel at multi-step reasoning, proper JSON formatting, and autonomo
 - **Large Local Models**: Some large local models (e.g., Qwen 3 32B, Llama 3.3 70B) can work with Native Mode, but results vary significantly by model quality.
 - **Small Local Models Warning**: **Small local models** (under 30B parameters) often struggle with Native Mode. They may produce malformed JSON, fail to follow strict state management, or make poor tool selection decisions. For these models, **Default Mode** is usually more reliable.
 
+#### Known Model-Specific Issues
+
+:::caution DeepSeek V3.2 Function Calling Issues
+**DeepSeek V3.2** has known issues with native function calling that cause **reproducible failures**. Despite being a 600B+ parameter model, it often outputs malformed tool calls.
+
+**The Problem**: DeepSeek V3.2 was trained using a proprietary format called **DSML (DeepSeek Markup Language)** for tool calls. When using native function calling, the model sometimes outputs raw DSML/XML-like syntax instead of proper JSON:
+- `<functionInvoke name="fetch_url">` instead of valid JSON
+- `<function_calls>` / `</function_calls>` tags in content
+- Garbled hybrid text like `prominentfunction_cinvoke name="search_parameter`
+
+**Why it happens**: This is heavily **model-dependent behavior induced during DeepSeek's fine-tuning process**. DeepSeek chose to train their model on DSML rather than standard OpenAI-style JSON tool calls. While inference providers (VertexAI, OpenRouter, etc.) attempt to intercept DSML blocks and convert them to OpenAI-style JSON, this translation layer is unreliable under certain conditions (streaming, high temperature, high concurrency, multi-turn conversations). **The primary responsibility lies with DeepSeek** for using a non-standard format that requires fragile translation.
+
+**Known contributing factors**:
+- Higher temperature values correlate with more malformed output
+- Multi-round conversations (6-8+ turns) can cause the model to stop calling functions entirely
+- Complex multi-step workflows (15-30 tool calls) may cause "schema drift" where argument formats degrade
+
+**Workarounds**:
+- **Use Default Mode** (prompt-based) instead of Native Mode for DeepSeek — this is the recommended approach
+- Lower temperature when using tool calling
+- Limit multi-round tool calling sessions
+- Consider alternative models for agentic workflows
+
+**This is a DeepSeek model/API issue**, not an Open WebUI issue. Open WebUI correctly sends tools in standard OpenAI format — the malformed output originates from DeepSeek's non-standard internal format.
+:::
+
 | Feature | Default Mode | Native Mode |
 |:---|:---|:---|
 | **Latency** | Medium/High | Low |

docs/features/workspace/knowledge.md

@@ -11,7 +11,7 @@ Knowledge part of Open WebUI is like a memory bank that makes your interactions
 - It's like a memory system for Open WebUI that allows it to pull from saved data, making responses more personalized and contextually aware.
 - You can use Knowledge directly in your chats with Open WebUI to access the stored data whenever you need it.
 
-Setting up Knowledge is straightforward! Simply head to the Knowledge section inside work space and start adding details or data. You don't need coding expertise or technical setup; it's built into the core system!
+Setting up Knowledge is straightforward! Simply navigate to **Workspace → Knowledge** in the sidebar and start adding details or data. You don't need coding expertise or technical setup; it's built into the core system!
 
 ## What is the "Knowledge" Section?
 
@@ -62,7 +62,7 @@ For complete details on all built-in agentic tools and how to configure them, se
 
 ### Setting Up Your Knowledge Base
 
-1. **Navigate to the Knowledge Section**: This area is designed to be user-friendly and intuitive.
+1. **Navigate to the Knowledge Section**: Click **Workspace** in the sidebar, then select **Knowledge**. This area is designed to be user-friendly and intuitive.
 2. **Add Entries**: Input information you want Open WebUI to remember. It can be as specific or broad as you like.
 3. **Save and Apply**: Once saved, the Knowledge is accessible and ready to enhance your chat interactions.
 

docs/getting-started/env-configuration.mdx

@@ -2997,9 +2997,22 @@ Strictly return in JSON format:
 
 - Type: `bool`
 - Default: `False`
-- Description: Enables or disables local web fetch for RAG.
+- Description: Controls whether RAG web fetch operations can access URLs that resolve to private/local network IP addresses.
 - Persistence: This environment variable is a `PersistentConfig` variable.
 
+When disabled (default), Open WebUI blocks web fetch requests to URLs that resolve to private IP addresses, including:
+
+- IPv4 private ranges (`10.x.x.x`, `172.16.x.x`-`172.31.x.x`, `192.168.x.x`, `127.x.x.x`)
+- IPv6 private ranges
+
+This is a **Server-Side Request Forgery (SSRF) protection**. Without this safeguard, a malicious user could provide URLs that appear external but resolve to internal addresses, potentially exposing internal services, cloud metadata endpoints, or other sensitive resources.
+
+:::warning
+
+Only enable this setting if you need to fetch content from internal network resources (e.g., an internal wiki or intranet) **and** you trust all users with access to your Open WebUI instance. Enabling this in a multi-tenant or public-facing deployment introduces significant security risk.
+
+:::
+
 ### Google Drive
 
 #### `ENABLE_GOOGLE_DRIVE_INTEGRATION`

docusaurus.config.ts

@@ -78,7 +78,7 @@ const config: Config = {
 			},
 			items: [
 				{
-					to: "blog",
+					href: "https://openwebui.com/blog",
 					label: "Blog",
 					position: "left",
 				},