add doc

Author: liqul

website/blog/authors.yml

@@ -0,0 +1,11 @@
+liqli:
+  name: Liqun Li
+  url: https://liqul.github.io
+  title: Principal Researcher
+  image_url: https://liqul.github.io/assets/logo_small_bw.png
+
+xu:
+  name: Xu Zhang
+  url: https://scholar.google.com/citations?user=bqXdMMMAAAAJ&hl=zh-CN
+  title: Senior Researcher
+  image_url: https://scholar.googleusercontent.com/citations?view_op=view_photo&user=bqXdMMMAAAAJ&citpid=3

website/blog/evaluation.md

@@ -1,4 +1,8 @@
-# How to evaluate a LLM agent?
+---
+title: How to evaluate a LLM agent?  
+authors: [liqli, xu]
+date: 2024-05-07  
+---
 
 ## The challenges
 It is nontrivial to evaluate the performance of a LLM agent. 

website/blog/experience.md

@@ -1,4 +1,8 @@
-# Experience selection
+---
+title: Experience Selection in TaskWeaver
+authors: liqli
+date: 2024-09-14
+---
 
 We have introduced the motivation of the `experience` module in [Experience](/docs/customization/experience) 
 and how to create a handcrafted experience in [Handcrafted Experience](/docs/customization/experience/handcrafted_experience).

website/blog/local_llm.md

@@ -1,4 +1,8 @@
-# Run TaskWeaver with Locally Deployed Not-that-Large Language Models
+---
+title: Run TaskWeaver with Locally Deployed Not-that-Large Language Models
+authors: liqli
+date: 2024-07-08
+---
 
 :::info
 The feature introduced in this blog post can cause incompatibility issue with the previous version of TaskWeaver

website/blog/plugin.md

@@ -1,4 +1,8 @@
-# Plugins In-Depth
+---
+title: Plugins In-Depth
+authors: liqli
+date: 2024-05-23
+---
 
 _**Pre-requisites**: Please refer to the [Introduction](/docs/plugin/plugin_intro) and the [Plugin Development](/docs/plugin/how_to_develop_a_new_plugin) 
 pages for a better understanding of the plugin concept and its development process._

website/blog/reasoning.md

@@ -1,4 +1,8 @@
-# What makes a good agent reasoning framework?
+---
+title: What makes a good agent reasoning framework?
+authors: liqli
+date: 2025-01-20
+---
 
 An agent can listen to the user's request, understand the context, make plans, take actions, observe the results, and respond to the user. Its behavior is driven by the reasoning process, which is the core of the agent's intelligence. 
 

website/blog/role.md

@@ -1,4 +1,8 @@
-# Roles in TaskWeaver
+---
+title: Roles in TaskWeaver
+authors: [liqli, xu]
+date: 2024-05-07
+---
 
 
 We frame TaskWeaver as a **code-first** agent framework. The term "code-first" means that the agent is designed to

website/blog/vision.md

@@ -0,0 +1,57 @@
+---
+title: Support vision input for the Planner
+authors: liqli
+date: 2025-03-13
+---
+
+## Introduction
+
+We have supported vision input for the Planner role in TaskWeaver. 
+The Planner role is responsible for generating the high-level plan for the task. 
+The vision input is a new type of input that contains images. 
+This feature is useful when the task requires visual understanding.
+<!-- truncate -->
+
+## How vision input is supported in TaskWeaver
+
+In TaskWeaver, we added a new role called `ImageReader` to read images and provide the image url (for remote images) or 
+the image encoded in base64 (for local images) to the Planner role.
+To have this new role, you need to include it in the project configure file as follows:
+
+```json
+{
+    "session.roles": [
+        "planner",
+        "code_interpreter",
+        "image_reader"
+    ]
+}
+```
+
+The ImageReader role takes the path or the url of the image as input and prepares a response Post for the Planner role. As described [here](https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/gpt-with-vision?tabs=rest) for Azure OpenAI API, if the image is local, ImageReader need to encode the image in base64 and pass it to the API. If the image is remote, ImageReader need to provide the url of the image. 
+The Planner role can then use the image information for various tasks.
+
+## Example
+
+Let's ask the agent to describe any uploaded image.
+
+
+![image_reader](../static/img/image_reader.png)
+
+The flow of the conversation is as follows:
+```mermaid
+graph TD
+    User --image path--> Planner
+    Planner --image path--> ImageReader
+    ImageReader --image encoded in Base64--> Planner
+    Planner --response--> User
+```
+
+In the example above, the User talks to the agent in Web UI and uploads an image. 
+TaskWeaver also support providing the image path in console mode, either using the `/load` command or just include 
+the image path in the input message.
+
+
+
+
+

website/docusaurus.config.js

@@ -58,6 +58,8 @@ const config = {
           // Remove this to remove the "edit this page" links.
           editUrl:
             'https://github.com/microsoft/TaskWeaver/tree/main/website',
+          blogSidebarTitle: 'All posts',
+          blogSidebarCount: 'ALL',
         },
         theme: {
           customCss: './src/css/custom.css',