gptme

Basic information

Website: https://web.archive.org/web/20241226055506/https://github.com/ErikBjare/gptme

Short description: “Personal AI assistant in your terminal, with tools so it can: Use the terminal, run code, edit files, browse the web, use vision, and much more; Assists in all kinds of knowledge-work, especially programming, from a simple but powerful CLI [command-line interface].” [source]

Intended uses: What does the developer say it’s for? Use Cases. (1) Write and run code faster with AI assistance. (2) Get the right command using natural language. (3) Process and analyze data directly in your terminal. (4) Experiment with new technologies or codebases hands-on. (5) Experiment with agents and tools in a local environment [e.g., give the assistant access to a full desktop, allowing it to interact with GUI applications — or — create your own agent with persistence using gptme-agent-template] [source]

Date(s) deployed: Main announcement on September 5, 2023 [source].

Developer

Website: https://web.archive.org/web/20241128133207/https://gptme.org/

Legal name: Erik Bjäreholt [source]

Entity type: Individual [source]

Country (location of developer or first author’s first affiliation): Sweden [source]

Safety policies: What safety and/or responsibility policies are in place? None

System components

Backend model: What model(s) are used to power the system? Variable, defaulting to GPT-4o [source]

Publicly available model specification: Is there formal documentation on the system’s intended uses and how it is designed to behave in them? None

Reasoning, planning, and memory implementation: How does the system ‘think’? gptme implements a simple loop for allowing the model to plan and act using default prompts that are available in documentation [source]

Observation space: What is the system able to observe while ‘thinking’? Gptme-agent can access terminal outputs, web pages, screenshots of your desktop, self-correction feedback, GUI applications, and files from a codebase [source]

Action space/tools: What direct actions can the system take? Examples of tools available to gptme include: Shell – Execute shell commands; Python – Run Python code; Browser – Browse and interact with web content; Vision – Process and analyze images [source]

User interface: How do users interact with the system? Besides the command-line interface, gptme can be used through a web-based interface [source] and as a GitHub bot [source].

Development cost and compute: What is known about the development costs? Unknown

Guardrails and oversight

Accessibility of components:

Weights: Are model parameters available? N/A; backends external model(s) via API
Data: Is data available? N/A; backends external model(s) via API
Code: Is code available? Available [source]
Scaffolding: Is system scaffolding available? Available [source]
Documentation: Is documentation available? Available [source]

Controls and guardrails: What notable methods are used to protect against harmful actions? Unknown

Customer and usage restrictions: Are there know-your-customer measures or other restrictions on customers? None

Monitoring and shutdown procedures: Are there any notable methods or protocols that allow for the system to be shut down if it is observed to behave harmfully? Unknown

Evaluation

Notable benchmark evaluations: None

Bespoke testing: Demos [source] and bespoke benchmarks [source].

Safety: Have safety evaluations been conducted by the developers? What were the results? None

Publicly reported external red-teaming or comparable auditing:

Personnel: Who were the red-teamers/auditors? None
Scope, scale, access, and methods: What access did red-teamers/auditors have and what actions did they take? None
Findings: What did the red-teamers/auditors conclude? None

Ecosystem information

Interoperability with other systems: What tools or integrations are available? As an open-source model, gptme is relatively interoperable with other systems. By default, it can be integrated with some editors, such as Vim [source]

Usage statistics and patterns: Are there any notable observations about usage? Unknown

Additional notes

None