Skip to content

Files

Latest commit

 

History

History
374 lines (258 loc) · 10.6 KB

README.md

File metadata and controls

374 lines (258 loc) · 10.6 KB

Set up Docs Agent CLI

This guide provides instructions on setting up Docs Agent's command-line interface (CLI) on your host machine for running Docs Agent tasks.

Docs Agent's agent runtask command allows you to run pre-defined chains of prompts, which are referred to as tasks. These tasks simplify complex interactions by defining a series of steps that the Docs Agent will execute. The tasks are defined in .yaml files stored in the tasks directory of your Docs Agent project. The tasks are designed to be reusable and can be used to automate common workflows, such as generating release notes, updating documentation, or analyzing complex information.

To set up the Docs Agent CLI, the steps are:

  1. Prerequisites
  2. Update your host machine's environment
  3. Clone the Docs Agent project repository
  4. Try the Docs Agent CLI

1. Prerequisites

Setting up Docs Agent requires the following prerequisite items:

  • A Linux host machine

  • A Google Cloud project with an API key enabled with the Generative Language API (that is, the Gemini API)

2. Update your host machine's environment

  1. Update the Linux package repositories on the host machine:

    sudo apt update

  2. Install the following dependencies:

    sudo apt install git pipx python3-venv

  3. Install poetry:

    pipx install poetry

  4. To add $HOME/.local/bin to your PATH variable, run the following command:

    pipx ensurepath

  5. To set the Google API key as a environment variable, add the following line to your $HOME/.bashrc file:

    export GOOGLE_API_KEY=<YOUR_API_KEY_HERE>

    Replace <YOUR_API_KEY_HERE> with the API key to the Gemini API.

  6. Update your environment:

    source ~/.bashrc

3. Clone the Docs Agent project

Note: This guide assumes that you're creating a new project directory from your $HOME directory.

  1. Clone the following repo:

    git clone https://github.com/google/generative-ai-docs.git

  2. Go to the Docs Agent project directory:

    cd generative-ai-docs/examples/gemini/python/docs-agent

  3. Install dependencies using poetry:

    poetry install

4. Try the Docs Agent CLI

  1. Enter the poetry shell environment:

    poetry shell

    Important: You must always enter the poetry shell environment to run the agent command.

  2. Enable autocomplete for Docs Agent CLI options in your environment:

    source scripts/autocomplete.sh

  3. Run the agent helpme command, for example:

    agent helpme how do I cook pasta?

    This command returns the Gemini model's response of your input prompt how do I cook pasta?.

  4. View the list of Docs Agent tasks available in your setup:

    agent runtask

    This command prints a list of Docs Agent tasks that you can run. (See the tasks directory in your local Docs Agent setup.)

  5. Run the agent runtask command, for example:

    agent runtask --task IndexPageGenerator

For more details on these commands, see the Interacting with language models section in the CLI reference page.

Appendices

Authorize credentials for Docs Agent

Note: This step may not be necessary if you already have OAuth client credentials (via gcloud) stored on your host machine.

This step is only necessary if you plan on using the agent tellme command to interact with your online corpora on Google Cloud.

Do the following:

  1. Download the client_secret.json file from your Google Cloud project.

  2. Copy the client_secret.json file to your host machine.

  3. Install the Google Cloud SDK on your host machine:

    sudo apt install google-cloud-sdk

  4. To authenticate credentials, run the following command in the directory of the host machine where the client_secret.json file is located:

    gcloud auth application-default login --client-id-file=client_secret.json --scopes='https://www.googleapis.com/auth/cloud-platform,https://www.googleapis.com/auth/generative-language.retriever'

    This command opens a browser and asks to log in using your Google account.

  5. Follow the instructions on the browser and click Allow to authenticate.

    This saves the authenticated credentials for Docs Agent (application_default_credentials.json) in the $HOME/.config/gcloud/ directory of your host machine.

Set up an alias to the gemini command

This section provides instructions on setting up the Docs Agent CLI to enable you to ask questions from anywhere in a terminal.

Using Docs Agent, you can configure your host machine's environment to make the gemini command run from anywhere in your terminal. The gemini command (which is an alias to Docs Agent's agent tellme command) reads a question from the arguments, asks the Gemini AQA model, and prints its response in the terminal.

The example below shows that a user can run the gemini command directly from a terminal:

user@user01:~$ gemini does Flutter support material design 3?

As of the Flutter 3.16 release, Material 3 is enabled by default.

To verify this information, see:

 • https://docs.flutter.dev/ui/design/material/index#more-information

user@user01:~$

In this setup, Docs Agent's AQA model is configured to use an example online corpus. However, using the tools available in the Docs Agent project, you can create and populate a new corpus with your own documents and adjust your Docs Agent configuration to use that corpus instead – you can also share the corpus with other members in your team.

To update your shell environment so that the gemini command can be run from anywhere in the terminal, do the following:

Note: If your Docs Agent project is not cloned in the $HOME directory, you need to edit the scripts/tellme.sh script in your docs-agent project directory.

  1. (Optional) Open the scripts/tellme.sh file using a text editor, for example:

    nano scripts/tellme.sh

    If necessary, adjust the path ($HOME/docs-agent) to match your docs-agent project directory on the host machine:

    # IF NECESSARY, ADJUST THIS PATH TO YOUR `docs-agent` DIRECTORY.
docs_agent_dir="$HOME/docs-agent"

    Save the file and close the text editor.

  2. Add the following alias line to your $HOME/.bashrc file:

    alias gemini='$HOME/docs-agent/scripts/tellme.sh'

    Similarly, if necessary, you need to adjust the path ($HOME/docs-agent) to match your the docs-agent project directory on the host machine.

  3. Update your environment:

    source ~/.bashrc

  4. Now you can run the gemini command from anywhere in your terminal:

    gemini <QUESTION>

    For example:

    user@user01:~/temp$ gemini does flutter support material design 3?

Set up your terminal to run the helpme command

Note: This is an experimental setup.

This new feature allows you to freely navigate a codebase setup in your terminal and asks Gemini to perform various tasks while automatically referencing the output you see in your terminal.

Similar to the agent tellme command, the agent helpme command allows you to ask a question to Gemini directly from your terminal. However, unlike the tellme command, the helpme command uses the Gemini Pro model and doesn't depend on an online corpus to retrieve relevant context. Instead, this helpme setup can read directly from the output of your terminal (that is, the last 150 lines at the moment) and automatically adds it as context to your prompt.

These tasks include, but not limited to:

  • Rewrite README file to be instructional and linear.
  • Rewrite README file to be more concise and better structured.
  • Format README to collect reference links at the bottom.
  • Write a protocol description.
  • Write comments for a C++ source file.

Note: Since this setup uses the Gemini Pro model, setting up OAuth on your host machine is not required.

To set up this helpme command in your terminal, do the following:

  1. (Optional) Open the scripts/helpme.sh file using a text editor, for example:

    nano scripts/helpme.sh

    If necessary, adjust the path ($HOME/docs-agent) to match your docs-agent project directory on the host machine:

    # IF NECESSARY, ADJUST THIS PATH TO YOUR `docs-agent` DIRECTORY.
docs_agent_dir="$HOME/docs-agent"

    Save the file and close the text editor.

  2. Add the following alias lines to your $HOME/.bashrc file:

    alias gemini-pro='$HOME/docs-agent/scripts/helpme.sh'
alias start_agent='script -f -o 200MiB -O /tmp/docs_agent_console_input'
alias stop_agent='exit'

    Similarly, if necessary, you need to adjust the path ($HOME/docs-agent) to match your the docs-agent project directory on the host machine.

  3. Update your environment:

    source ~/.bashrc

  4. When you are ready to let Docs Agent to read output from your terminal, run the following command:

    start_agent

    Note: To stop this process, run stop_agent.

  5. Navigate to a directory in your terminal and use the cat command (or head or tail) to print the content of a file to your terminal.

    (In fact, you can run any command that prints output to the terminal.)

    For example:

    user@user01:~/my-example-project$ cat test.cc
<prints the test.cc file here>

  6. To use the latest output from your terminal, run the gemini-pro command immediately after the output:

    gemini-pro <REQUEST>

    For example:

    user@user01:~/my-example-project$ cat test.cc
<prints the test.cc file here>
user@user01:~/my-example-project$ gemini-pro could you help me write comments for this C++ file above?