Skip to main content

npcsh is a command line tool for integrating LLMs into everyday workflows

Project description

npcsh

Welcome to npcsh, the shell for interacting with NPCs (LLM-powered AI agents) and for coordinating actions and information between the NPCs.

npcsh is meant to be a drop-in replacement shell for any kind of bash/zsh/powershell and allows the user to directly operate their machine through the use of the LLM-powered shell.

npcsh introduces a new paradigm of programming for LLMs: npcsh allows users to set up NPC profiles (a la npc_profile.npc) where a user sets the primary directive of the NPC, the tools they want the NPC to use, and other properties of the NPC. NPCs can interact with each other and their primary directives and properties make these relationships explicit through jinja references.

With npcsh, we can more seamlessly stick together complex workflows and data processing tasks to form NPC Assembly Lines where pieces of information are evaluated in a sequence by different NPCs and the results are passed along to the next NPC in the sequence.

Dependencies

  • ollama
  • python >3.10

The default model is currently phi3. Download it by running

ollama run phi3

We support inference as well via openai and anthropic. To use them, set an ".env" file up in the folder where you are working and set the API keys there.

Eventually, we will add the ability to use any huggingface model.

The user can change the default model by setting the environment variable NPCSH_MODEL in their ~/.npcshrc to the desired model name and to change the provider by setting the environment variable NPCSH_PROVIDER to the desired provider name.

The provider must be one of ['ollama', 'openai', 'anthropic'] and the model must be one available from those providers.

Linux install

sudo apt-get install espeak
sudo apt-get install portaudio19-dev python3-pyaudio
sudo apt-get install alsa-base alsa-utils
sudo apt-get install libcairo2-dev
sudo apt-get install libgirepository1.0-dev
sudo apt-get install ffmpeg
pip install npcsh

Mac install

brew install portaudio
brew install ffmpeg
brew install ollama
brew services start ollama
brew install pygobject3
pip install npcsh

Usage

After it has been pip installed, npcsh can be used as a command line tool. Start it by typing:

npcsh

Once in the npcsh, you can use bash commands or write natural language queries or commands. You can also switch between different modes defined below and you can compile a network of NPCs or use the macro tools we have developed.

compilation

Each NPC can be compiled to accomplish their primary directive and then any issues faced will be recorded and associated with the NPC so that it can reference it later through vector search. In any of the modes where a user requests input from an NPC, the NPC will include RAG search results before carrying out the request.

Base npcsh

In the base npcsh shell, inputs are processed by an LLM. The LLM first determines what kind of a request the user is making and decides which of the available tools or modes will best enable it to accomplish the request.

spool mode

Spool mode allows the users to have threaded conversations in the shell, i.e. conversations where context is retained over the course of several turns. Users can speak with specific NPCs in spool mode by doing /spool <npc_name> and can exit spool mode by doing /exit.

Built-in NPCs

Built-in NPCs are NPCs that should offer broad utility to the user and allow them to create more complicated NPCs. These built-in NPCs facilitate the carrying out of many common data processing tasks as well as the ability to run commands and to execute and test programs.

Bash NPC

The bash NPC is an NPC focused on running bash commands and scripts. The bash NPC can be used to run bash commands and the user can converse with the bash NPC by doing /spool bash to interrogate it about the commands it has run and the output it has produced. A user can enter bash mode by typing /bash and can exit bash mode by typing /bq. Use the Bash NPC in the profiles of other NPCs by referencing it like {{bash}}.

Command NPC

The LLM or specific NPC will take the user's request and try to write a command or a script to accomplish the task and then attempt to run it and to tweak it until it works or it's exceeded the number of retries (default=5).

Use the Command NPC by typing /cmd <command>. Chat with the Command NPC in spool mode by typing /spool cmd. Use the Command NPC in the profiles of other NPCs by referencing it like {{cmd}}.

Question NPC

The user can submit a 1-shot question to a general LLM or to a specific NPC. Use it like /question <question> <npc_name> or /question <question>

You can also chat with the Question NPC in spool mode by typing /spool question.

Other useful tools

Over-the-shoulder

Over the shoulder allows the user to select an area of the screen and the area will be passed to a vision LLM and then the user can inquire about the image or ask for help with it.

data

load data in, manipulate it, analyze it.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

npcsh-0.1.18.tar.gz (38.5 kB view details)

Uploaded Source

Built Distribution

npcsh-0.1.18-py3-none-any.whl (38.0 kB view details)

Uploaded Python 3

File details

Details for the file npcsh-0.1.18.tar.gz.

File metadata

  • Download URL: npcsh-0.1.18.tar.gz
  • Upload date:
  • Size: 38.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/5.1.1 CPython/3.9.20

File hashes

Hashes for npcsh-0.1.18.tar.gz
Algorithm Hash digest
SHA256 d620bea92b7e8d21ce338e97efbc2540a1d4725286003676d36a8240c57c4a83
MD5 0ffc7b4f126c12d6481067107d5b0ab2
BLAKE2b-256 01322279aca4b5aae8c6312cda3f50abc75627f198513d0938399d4d118ab68b

See more details on using hashes here.

File details

Details for the file npcsh-0.1.18-py3-none-any.whl.

File metadata

  • Download URL: npcsh-0.1.18-py3-none-any.whl
  • Upload date:
  • Size: 38.0 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/5.1.1 CPython/3.9.20

File hashes

Hashes for npcsh-0.1.18-py3-none-any.whl
Algorithm Hash digest
SHA256 1bfd4f4e5f52782775fbda623957eaa38f0e48514b336f580c2492bfc546626e
MD5 115be5cbbf86a0d9a46b8417dba05ad3
BLAKE2b-256 5813f6a7ff58c8f7ee313de2676363ebd04314aa5dfb58a4139d1832a55ed258

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page