MicrosoftDocs · Taojunshen · May 16, 2025 · May 16, 2025
@@ -1,7 +1,7 @@
 ---
 title: How to create an agent for Ollama
 description: Learn how to create an agent for the Ollama language model in AI Shell.
-ms.date: 12/12/2024
+ms.date: 05/16/2025
 ms.topic: how-to
 ---
 
@@ -15,21 +15,24 @@ get the desired output or assistance. Agents are implemented as C# classes that
 For details about the `AIShell.Abstraction` layer and `AIShell.Kernel`, see the
 [AI Shell architecture][05] documentation.
 
+This article is a step-by-step guide to creating an agent for the [Ollama][01] language model. The
+purpose of this article is to provide a simple example of how to create an agent. There's a more
+robust implementation of the Ollama agent in the [`AIShell.Ollama.Agent`][04] folder of the
+repository.
+
 ## Prerequisites
 
 - .NET 8 SDK or newer
 - PowerShell 7.4.6 or newer
 
 ## Steps to create an agent
 
-For this example we create an agent to communicate with the language model `phi3` by utilizing
-[Ollama][01], a CLI tool for managing and using locally built LLM/SLMs. The complete source code of
-the agent can be found in the [`shell/agents/AIShell.Ollama.Agent`][04] folder of the repository.
-the repository.
+For this example, we create an agent to communicate with the language model `phi3` using
+[Ollama][01]. Ollama is a CLI tool for managing and using locally built LLM/SLMs.
 
 ### Step 1: Create a new project
 
-First step is to create a new classlib project.
+First step is to create a new **classlib** project.
 
 1. Create a new folder named `OllamaAgent`
 1. Run the following command to create a new project:
@@ -94,8 +97,7 @@ public sealed class OllamaAgent : ILLMAgent
 
 Next, implement the necessary variables and methods of the agent class. The comments provide
 descriptions of the members of the **OllamaAgent** class. The `_chatService` member is an instance
-of the **OllamaChatService** class. You will implement the **OllamaChatService** class in a later
-step.
+of the **OllamaChatService** class, which you implement in a later step.
 
 ```csharp
 public sealed class OllamaAgent : ILLMAgent
@@ -204,8 +206,8 @@ public sealed class OllamaAgent : ILLMAgent
 }
 ```
 
-For the initial implementation, the agent to returns "Hello World!" to prove that you created the
-correct interfaces. You will also add a `try-catch` block to catch and handle any exceptions when
+For the initial implementation, the agent to returns "Hello World!", proving that you created the
+correct interfaces. You also need to add a `try-catch` block to catch and handle any exceptions when
 the user tries to cancel the operation.
 
 Add the following code to your `Chat` method.
@@ -267,7 +269,7 @@ Before you can use the Ollama API, you need to create classes that send input to
 responses from the Ollama API. The following [Ollama example][02] shows the format of the input and
 the response from the agent.
 
-For this example we call the Ollama API with streaming disabled. Ollama generates a single, fixed
+This example calls the Ollama API with streaming disabled. Ollama generates a single, fixed
 response. In the future, you could add streaming capabilities so that responses could be rendered in
 real time, as the agent receives them.
 
@@ -465,7 +467,7 @@ This command builds all necessary packages in the `\bin\Debug\net8.0` folder of
 To have `aish` load the agent, you need to copy the `.dll` files to a folder in the `Agents` folder.
 The folder name should be the same as the agent name.
 
-There are two location where you can install agents:
+You can install agents in one of two locations:
 
 - In the `Agents` folder under the location where you installed `aish.exe`. The
   [install script][08] for AI Shell installs in `%LOCALAPPDATA%\Programs\AIShell`. Create the
@@ -488,10 +490,11 @@ Please select an agent to use:
 
 ## How can I share my own agent?
 
-Currently there is no way to share your agents in a centralized repository. We suggest forking this
-repository for development of your own agent. You can share a link your fork in the `Agent Sharing`
-section of the [Discussions][03] tab of this repository. To use an agent, if you put its `dll` files
-in the `agents` folder of the base directory of `aish.exe`, the agent will be loaded by `aish`.
+There's no way to share your agents in a centralized repository. We suggest forking this repository
+for development of your own agent. You can share a link your fork in the `Agent Sharing` section of
+the [Discussions][03] tab of this repository. To use an agent, put agent `dll` files in the `agents`
+folder of the base directory of `aish.exe`. AI Shell automatically loads the agents from that
+folder.
 
 <!-- link references -->
 [01]: https://github.com/ollama/ollama

@@ -15,7 +15,7 @@ You can also use this example code as a template to create your own agent.
 
 - Install [Ollama][01]
 - Install a [Ollama model][02], we
-  suggest using the `phi3` model as it is set as the default model in the code
+  suggest using the `phi3` model as it's set as the default model in the code
 - [Start the Ollama API server][03]
 
 ## Configuration
@@ -35,8 +35,6 @@ There is an updated version of the Ollama agent available in the AI Shell reposi
   queries
 - Streaming is currently not supported if you change the stream value to `true` in the data to send
   to the API it will not work
-- Configuration is currently hard coded in the code and will be moved to a configuration file in the
-  future
 
 <!-- link references -->
 [01]: https://github.com/ollama/ollama

@@ -1,7 +1,7 @@
 ---
 title: OpenAI agent README
 description: Learn how to use the OpenAI agent in AI Shell.
-ms.date: 03/03/2025
+ms.date: 05/16/2025
 ms.topic: how-to
 ---
 # OpenAI agent
@@ -23,9 +23,14 @@ recommend using an Azure OpenAI deployment for enhanced security and privacy.
 
 ## Configuration
 
-Before getting started you need to configure the agent with the details of your OpenAI details. To
-configure the agent, run `/agent config openai-gpt` to open up the setting file in your default
-editor. Update the file based on the following example:
+Before getting started, you need to configure the agent with the details of your OpenAI
+implementation. To configure the agent, use the `/agent config openai-gpt` command to open
+configuration file in your default editor.
+
+- Azure OpenAI requires the `endpoint`, `deployment` name, and the `model` name.
+- Public OpenAI only requires the `model` name. No endpoint, no deployment name.
+
+Update the file based on the following example:
 
 ```jsonc
 {
@@ -68,7 +73,7 @@ editor. Update the file based on the following example:
 ```
 
 > [!NOTE]
-> The endpoint for the Azure OpenAI configuration does not need a full endpoint including the
+> The endpoint for the Azure OpenAI configuration doesn't need a full endpoint including the
 > deployment, for example you can just use the following endpoint format,
 > `https://<YourServiceName>.openai.azure.com`.
 
@@ -85,8 +90,8 @@ select different OpenAI models for each GPT as required.
 
 The command `/gpt` is provided to make it easy to manage the GPTs.
 
-- Run `/gpt use <gpt-name>` to switch to another GPT instance, or run `/gpt use` to simply choose
-  from the available ones.
+- Run `/gpt use <gpt-name>` to switch to another GPT instance, or run `/gpt use` to choose from the
+  available ones.
 - Run `/gpt list <gpt-name>` to view the details of a GPT definition, or run `/gpt list` to list all
   available GPTs.
 
@@ -108,10 +113,10 @@ Commands:
 
 ## Support for other OpenAI-compatible models
 
-With the release of AI Shell v1.0.0-preview.2, the `openai-gpt` agent extends support to third party
+With the release of AI Shell v1.0.0-preview.2, the `openai-gpt` agent extends support to third-party
 models that follow the OpenAI API specifications, allowing for a more flexible AI experience. Many
-of these models are open source tools for running SLMs and LLMs locally. The `openai-gpt` agent nows
-supports the following additional models:
+of these models are open source tools for running SLMs and LLMs locally. The `openai-gpt` agent now
+supports the following models:
 
 - [**Ollama**][08]
 - [**LM Studio**][06]
@@ -120,10 +125,23 @@ supports the following additional models:
 - [**Google Gemini**][02]
 - [**Grok**][05]
 
-To use these models, you only need to configure the `endpoint`, `key` and `model name` in the agent
-config file. The following `openai.agents.config` file contains example configurations for two of
-the new AI providers. For more information about endpoints and model names, see to the documentation
-for the specific model.
+To use these models, you only need to configure the `endpoint`, `key`, and `model` name in the agent
+configuration file. The following `openai.agents.config` file contains example configurations for
+two of the new AI providers. For more information about endpoints and model names, see the
+documentation for the specific model.
+
+With the release of AI Shell v1.0.0-preview.4, the `openai-gpt` agent supports the following
+models names:
+
+- `o1`
+- `o3`
+- `o4-mini`
+- `gpt-4.1`
+- `gpt-4o`
+- `gpt-4`
+- `gpt-4-32k`
+- `gpt-4-turbo`
+- `gpt-3.5-turbo`
 
 <!-- link references -->
 [01]: /azure/ai-services/openai/how-to/create-resource?pivots=web-portal

@@ -1,7 +1,7 @@
 ---
 title: AI Shell command reference
 description: Learn about the command-line options and commands available in AI Shell.
-ms.date: 10/29/2024
+ms.date: 05/16/2025
 ms.topic: reference
 ---
 # AI Shell command reference
@@ -33,21 +33,35 @@ By default, `aish` provides a base set of chat commands used to interact with th
 list of commands, use the `/help` command in the chat session.
 
 ```
-  Name       Description                                      Source
-──────────────────────────────────────────────────────────────────────
-  /agent     Command for agent management.                    Core
-  /cls       Clear the screen.                                Core
-  /code      Command to interact with the code generated.     Core
-  /dislike   Dislike the last response and send feedback.     Core
-  /exit      Exit the interactive session.                    Core
-  /help      Show all available commands.                     Core
-  /like      Like the last response and send feedback.        Core
-  /refresh   Refresh the chat session.                        Core
-  /render    Render a markdown file, for diagnosis purpose.   Core
-  /retry     Regenerate a new response for the last query.    Core
+  Name       Description                                                 Source
+─────────────────────────────────────────────────────────────────────────────────────
+  /agent     Command for agent management.                               Core
+  /clear     Clear the screen.                                           Core
+  /code      Command to interact with the code generated.                Core
+  /dislike   Dislike the last response and send feedback.                Core
+  /exit      Exit the interactive session.                               Core
+  /help      Show all available commands.                                Core
+  /like      Like the last response and send feedback.                   Core
+  /refresh   Start a new chat session.                                   Core
+  /retry     Regenerate a new response for the last query.               Core
 ```
 
-## `/agent` sub commands
+## `/agent`
+
+Command for agent management.
+
+```
+agent [command] [options]
+```
+
+Options: -h, --help  Show help and usage information
+
+Subcommands
+
+- `config <azure|openai-gpt>` - Open up the setting file for an agent. When no agent is specified,
+  target the active agent.
+- `list` - List all available agents.
+- `use <azure|openai-gpt>` - Specify an agent to use, or choose one from the available agents.
 
 ### `/agent config`
 
@@ -62,7 +76,6 @@ Arguments: `<azure|openai-gpt>` Name of an agent.
 Options:
 
 - `--editor <editor>` The editor to open the setting file in.
-- `-h`, `--help` Show help and usage information
 
 Example:
 
@@ -74,10 +87,6 @@ Example:
 
 List all available agents.
 
-Options:
-
-- `-h`, `--help`  Show help and usage information
-
 Example:
 
 ```
@@ -99,14 +108,41 @@ Example:
                    2. Define the GPT(s). See details at
                       https://aka.ms/aish/openai
                    3. Run '/refresh' to apply the new settings.
-  azure (active)   This AI assistant can generate Azure CLI and
-                   Azure PowerShell commands for managing
-                   Azure resources, answer questions, and provides
-                   information tailored to your specific Azure
-                   environment.
+  azure            This AI assistant connects you to the Copilot in
+                     Azure and can generate Azure CLI and Azure
+                     PowerShell commands for managing Azure resources
+                     and answer questions about Azure.
+```
+
+### `/agent use`
+
+Specify an agent to use, or choose one from the available agents.
+
+```
+agent use [<agent>] [options]
+```
+
+Arguments: `<azure|openai-gpt>` - Name of an agent.
+
+## `/clear`
+
+Clears the screen. You can also use the alias `/cls`.
+
+## `/code`
+
+Command to interact with the code generated.
+
 ```
+code [command] [options]
+```
+
+Subcommands:
 
-## `/code` sub commands
+- `copy <n>` Copy the n-th (1-based) code snippet to clipboard. Copy all the code when `<n>` isn't
+  specified. [default: -1]
+- `save <file>` - Save all the code to a file.
+- `post <n>` - Post the n-th (1-based) code snippet to the connected command-line shell. Post all
+  the code when `<n>` isn't specified. [default: -1]
 
 ### `/code copy`
 
@@ -121,11 +157,7 @@ code copy [<n>] [options]
 Arguments:
 
 - `<n>` Use the n-th (1-based) code snippet. Use all the code when no value is specified.
-[default: -1]
-
-Options:
-
-- `-h`, `--help`  Show help and usage information
+  [default: -1]
 
 Examples
 

@@ -24,17 +24,23 @@ AI Shell is supported on the following platforms:
 - Windows 10 or higher
 - PowerShell 7.4.6 or higher
 - Windows Terminal
+- PSReadLine v2.4.2-beta2 or higher
 
 ### [macOS](#tab/macos)
 
 - macOS v13 Ventura or higher
 - PowerShell 7.4.6 or higher
-- iTerm2 preferred (no split-screen experience)
+- Default macOS Terminal app for the standalone experience
+- iTerm2 required to use the split-screen experience
+  - Python 3.11 or higher required to support iTerm2
+  - Must enable **Python APIs** in the iTerm2 settings
+- PSReadLine v2.4.2-beta2 or higher
 
 ### [Linux](#tab/linux)
 
 - Ubuntu 20.04 or higher
-- PowerShell 7.4 or higher (recommended)
+- PowerShell 7.4.6 or higher
+- PSReadLine v2.4.2-beta2 or higher
 - Any terminal emulator supported by the OS (no split-screen experience)
 
   > [!NOTE]
@@ -55,8 +61,8 @@ On Windows, this script:
 
 On macOS, this script:
 
-- Installs the `aish` executable to `/usr/local/AIShell`
-- Due to some limitations, the **AIShell** module is not installed
+- Installs the `aish` executable to `/usr/local/AIShell` and creates a symbolic link to
+  `/usr/local/bin/aish`.
 
 > [!NOTE]
 > This script only works on Windows and Mac systems. Linux users need to follow the manual