️
A versatile workflow automation system. Create, organize, and execute complex AI-driven tasks.
Agent Pilot provides a seamless experience, whether you want to chat with a single LLM or a complex multi-member workflow.
With an intuitive and feature-rich interface, users can effortlessly design AI workflows and chat with them in real-time.
Branching chats are supported, allowing flexible interactions and iterative refinement.
| Platform | Downloads |
|---|---|
| Linux |
Mirror: AgentPilot_0.3.2_Linux_Portable.tar.gz |
| Windows |
Mirror: AgentPilot_0.3.2_Windows_Portable.zip MD5: 034c1ecfda52ecdba6f560515e36232f SHA1: c2904d0adffd43421ce8498c90d5545758389904 |
Building from source: How to build from source
Tip
You can migrate your old database to the new version by replacing your executable with the new one before starting the application.
Create new agents, edit their configuration and organise them into folders.
Multi-member workflows can be saved as a single agent and nested infinitely.
View, continue and delete previous workflow chats and organise them into folders.
Messages, tools and code can be edited and re-run, allowing a more practical way to chat with your workflow.
Branching works with all plugins and multi-member chats.
Seamlessly add other members or blocks to a workflow and configure how they interact with each other.
Members aligned vertically are executed in parallel.
Available members:
- User - This is you and will await your input.
- Agent - Gets an LLM response with integrated tools and messages.
- Text - A simple text block that can nest other blocks.
- Code - Gets the output of any given code.
- Prompt - Gets an LLM response from a single prompt.
- Module - Runs or retrieves a method or variable from any module.
- Workflow - Any combination of the above types.
Manage a collection of nestable blocks available to use in any workflow or text field,
allowing reusability and consistency.
By default a block is a simple text block, but it can be any of the above member types, even a multi-member workflow.
These can be quickly dropped into any workflow, or used in text fields (such as system message) by using the block name in curly braces, e.g. {block-name}.
Create and manage tools which can be assigned to agents.
Tools share the same functionality as blocks, except by default they are a single Code member.
They can also be an entire workflow, this allows your agents to not only run code but an entire workflow if you wish.
Configure their parameters, which can be accesed from all workflow member types.
These parameters can be modified at runtime and re-executed, this creates a branch point which you can cycle through.
Modules are python files which are imported at runtime.
These are useful for things like toolkits, daemons, memory, custom pages or anything that needs persistence.
Includes a flexible and powerful set of base classes for building complex hierarchical configuration interfaces. The entire app is built on this framework. Developers can modify or create configuration pages easily, even while the app is running.
Open Interpreter is integrated into Agent Pilot, and can either be used standalone as a plugin or used to execute code in 9 languages (Python, Shell, AppleScript, HTML, JavaScript, PowerShell, R, React, Ruby)
Code can be executed in multiple ways:
- From any 'Code' member in any workflow (Chat, Block, Tool).
- From a message with the role 'Code'
You should always understand the code that is being run, any code you execute is your own responsibility.
For code messages, auto-run can be enabled in the settings. To see code messages in action talk to the pre-configured Open Interpreter agent.
Blocks under the 'System Blocks' folder are used for generating or enhancing fields. Claude's prompt generator is included by default, you can tweak it or create your own.
- Prompt - AI enhanced user input
- Agent - AI generated agent (Coming soon)
-
- System message - AI generated system message (Coming soon)
- Page - AI generated page (Coming soon)
Agent Pilot supports the following plugins:
-
Agent - Create custom agent behaviour.
-
- CrewAI Agent (Currently disabled)
-
Workflow - Create workflow behaviour.
-
- CrewAI Workflow (Currently disabled)
-
Provider - Add support for a model provider.
Coming back soon
Agents can be linked to a text-to-speech service, combine with a personality context block and make your agent come to life!
LiteLLM is integrated and supports the following providers:
- AI21
- AWS Bedrock
- AWS Sagemaker
- Aleph Alpha
- Anthropic
- Anyscale
- Azure OpenAI
- Baseten
- Cloudflare
- Cohere
- Custom API Servers
- DeepInfra
- DeepSeek
- Gemini
- Github
- Groq
- Huggingface
- Mistral
- NLP Cloud
- Nvidia NIM
- Ollama
- OpenAI
- OpenRouter
- PaLM API Google
- Perplexity AI
- Petals
- Replicate
- Together AI
- VLLM
- VertexAI Google
- Voyage
Contributions to Agent Pilot are welcome and appreciated. Please feel free to submit a pull request.
- Changing the config of an OpenAI Assistant won't reload the assistant, for now close and reopen the chat.
- Be careful using auto run code and open interpreter, any chat you open, if code is the last message it will start auto running, I'll add a flag to remember if the countdown has been stopped.
- Sometimes the scroll position jumps if the user is scrolled up and an AI response has finished generating.
- Windows exe must have console visible due to a strange bug.
- Issue on linux, creating venv does not install pip
- Numeric tool parameters get stuck at -99999
- When editing a previous message with markdown, to resend you have to press the resend button twice (because the first click makes the bubble lose focus, which blocks the event button click event)
- Multiple async groups (vertically aligned members) in the same workflow causes an issue.
If you find this project useful please consider showing support by giving a star or leaving a tip :)
BTC:
ETH: