Skip to content

Latest commit

 

History

History
75 lines (55 loc) · 1.66 KB

USAGE.md

File metadata and controls

75 lines (55 loc) · 1.66 KB
Raw

Usage

Bot CLI

The bot CLI is the main entry point for running the bot. It provides a few commands to run the bot in different environments and to increase the verbosity of the logs.

❯ sam run --help
Usage: sam run [OPTIONS] COMMAND1 [ARGS]... [COMMAND2 [ARGS]...]...

  Run an assistent bot.

Options:
  -v, --verbose  Enables verbose mode.
  --help         Show this message and exit.

Commands:
  slack  Run the Slack bot demon.

To run the Slack bot all you need is to run:

sam run slack

Training CLI

The training CLI is used to "train" the OpenAI assistant. You can upload the system prompts to OpenAI to align the assistant with your company's specific needs.

❯ sam assistants --help
Usage: sam assistants [OPTIONS] COMMAND1 [ARGS]... [COMMAND2 [ARGS]...]...

  Manage OpenAI assistants.

Options:
  -v, --verbose  Enables verbose mode.
  --help         Show this message and exit.

Commands:
  list    List all assistants configured in your project.
  upload  Compile and upload all assistants system prompts to OpenAI.

To upload the system prompts to OpenAI you can run:

sam assistants upload

System Prompts

You can configure assistants in your pyproject.toml file.

# pyproject.toml
[[tool.prompts.assistants]]
name = "Sam"
assistant_id = "asst_1234567890"
project = "default-project"
instructions = [
  "assistants/sam.md",
  "assistants/shared/values.md",
  "assistants/shared/principles.md",
  "assistants/shared/style.md",
]

The instructions field is a list of markdown files that contain the system prompts. You will need to set they OpenAI secret key in the environment variable.