Description
gptme is a personal AI agent designed to operate seamlessly within any terminal environment, from local laptops and SSH sessions to CI pipelines. It distinguishes itself with a provider-agnostic and local-first approach, ensuring user data privacy and model flexibility. The agent comes pre-equipped with essential tools for shell commands, Python execution, web browsing, and vision capabilities, making it a versatile assistant for a wide range of knowledge-work and coding tasks.
Read More
gptme supports a broad spectrum of LLM providers, including Anthropic, OpenAI, Google, xAI, and DeepSeek, as well as fully local models via llama.cpp. This flexibility allows users to select the most appropriate model for specific tasks, balancing cost and performance. It is positioned as a capable alternative to other AI coding assistants like Claude Code, Codex, Cursor, and Warp, being one of the pioneering agent CLIs available since Spring 2023 and remaining under active development.
The platform emphasizes extensibility through a layered system of plugins, skills, and lessons. Plugins allow for custom tools and commands, skills provide reusable workflow bundles, and lessons offer contextual guidance. gptme also features a robust ecosystem, including a web UI (gptme-webui), a template for building autonomous agents (gptme-agent-template), and community contributions like gptme-codegraph for advanced code analysis. Its integration capabilities extend to MCP (Model Context Protocol) and ACP (Agent Client Protocol), enabling it to interact with various servers and editors.
Beyond interactive use, gptme is engineered for autonomous agent functionality. The gptme-agent-template facilitates the creation of persistent agents with features like run loops, task management, and meta-learning. The reference implementation, 'Bob,' demonstrates continuous autonomous operation, handling tasks such as code reviews and task queue management. Guardrails are integrated to ensure safe and focused autonomous operation, including input, pre-action, and output checks.
Use cases span development, shell expertise, data analysis, interactive learning, research automation, and building autonomous agents. gptme is committed to being an extensible, well-tested, and clean codebase, welcoming community contributions. Future developments include a desktop app and a managed cloud service.
- CategoryAI Coding Assistants
- Pricing ModelOpen Source
- DeploymentLocal, Cloud (via gptme.ai WIP)
- PlatformsWeb
gptme's Core Features
Runs in any terminal environment (laptop, SSH, tmux, CI)
Provider-agnostic and local-first AI agent
Supports numerous LLM providers (Anthropic, OpenAI, Google, xAI, DeepSeek, OpenRouter, local)
Built-in tools for shell, Python, web browsing, vision, file manipulation
Extensible via plugins, skills, and lessons system
Enables autonomous agent operation with persistent workspaces and run loops
Offers a web UI (gptme-webui) for browser-based interaction
Integrates with MCP and ACP protocols for broader ecosystem compatibility
Includes advanced code intelligence with gptme-codegraph
Provides guardrails for autonomous agent safety and focus
Supports interactive and non-interactive modes for scripting and CI/CD
Features a self-correcting mechanism for improved responses
How to use gptme?
Install: Use pipx or uv to install gptme.
Configure: Set up LLM provider credentials or use OpenRouter.
Run: Execute 'gptme' in your terminal to start a conversation.
Interact: Type your requests, and gptme will respond using available tools.
Code: Use gptme for code generation, debugging, and refactoring.
Automate: Employ non-interactive mode for CI/CD pipelines or scripts.
Extend: Develop custom plugins or skills to tailor gptme to your workflow.
Autonomous Agents: Utilize gptme-agent-template to create persistent autonomous agents.
gptme's Use Cases
- Terminal Coding Assistant
- Shell Command Expert
- Data Analysis Tool
- Interactive Learning Platform
- Autonomous Agent Development
- Research Automation
- CI/CD Integration
- Knowledge Work Assistant
