smolagents is a library that enables you to run powerful agents in a few lines of code. The logic for agents fits in about 1,000 lines of code. Our CodeAgent writes its actions in code (as opposed to "agents being used to write code"). To make it secure, we support executing in sandboxed environments via E2B or via Docker. Supports any LLM. It can be a local transformers or ollama model, one of many providers on the Hub, or any model from OpenAI, Anthropic and many others via our LiteLLM integration. Agents support text, vision, video, even audio inputs!
Nerve is an ADK (Agent Development Kit) designed to be a simple yet powerful platform for creating and executing LLM-based agents. Agents are simple YAML files that can use a set of built-in tools such as a bash shell, file system primitives and other things (like APIs).
(archived) https://www.evilsocket.net/2025/03/13/How-To-Write-An-Agent/
Keychain helps you to manage SSH and GPG keys in a convenient and secure manner. It acts as a frontend to ssh-agent and ssh-add, but allows you to easily have one long running ssh-agent process per system, rather than the norm of one ssh-agent per login session.
This dramatically reduces the number of times you need to enter your passphrase. With keychain, you only need to enter a passphrase once every time your local machine is rebooted. Keychain also makes it easy for remote cron jobs to securely "hook in" to a long-running ssh-agent process, allowing your scripts to take advantage of key-based logins.
If your workflow is such that you have one terminal window but multiple tabs, this will fix quite a few annoyances.
Github: https://github.com/funtoo/keychain
Works very well on Linux, I've been using it for years. Also works on OSX; it's in Homebrew.
A de facto directory of third party Huginn agents that are built using huginn_agent as a template.
A skeleton for Huginn agent development. Specifically, for developing agents as Ruby gems and not native to the environment. Explains how to use it to stand up the skeleton of an agent for development, how to register it with a Huginn instance so it can be used, and how to run unit tests. Has a short list of agents that are built using this framework.
I wonder if this could be used to re-implement activeworkflow-agent-python or activeworkflow-remote-agent-api.
A Huginn agent that can read Google Calendars using the GCal API and emit events based upon what it finds.
Insight into the hidden ecosystem of autonomous chatbots and data scrapers crawling across the web. Protect your website from unwanted AI agent access. You can submit newly spotted agents. No data feeds so you have to sign up.
Maintaining a robots.txt file will help protect your website from unwanted AI agent access. The Dark Visitors list is continuously updated so you can control the behavior of all known AI agents.
The "Awesome GPTs (Agents) Repo" represents an initial effort to compile a comprehensive list of GPT agents focused on cybersecurity (offensive and defensive), created by the community. Please note, this repository is a community-driven project and may not list all existing GPT agents in cybersecurity. Contributions are welcome – feel free to add your own creations!
Disclaimer: Users should exercise caution and evaluate the agents before use. Additionally, please note that some of these GPTs are still in experimental test phase.
Welcome to our list of AI agents. We structured the list into two parts: Open source projects and closed-source projects and companies. The list is done according to our best knowledge, although definitely not comprehensive.
Willow is an ESP IDF based project primarily targetting the ESP BOX hardware from Espressif. Our goal is to provide Amazon Echo/Google Home competitive performance, accuracy, cost, and functionality with Home Assistant and other platforms - 100% open source and completely self-hosted by the user with "ready for the kitchen counter" low cost commercially available hardware.
Use Willow Inference Server anywhere or don't use it at all with command recognition on the device. Have the results go anywhere you want. Integrate with whatever you want. Completely open source so it does what you want, only what you want, and only how you want it. No more annoying extra prompts or sales pitches to upsell you. Supports multiple wake words with more coming soon.
Approximately $50 hardware cost (plus USB-C power supply). Fully assembled. Done.
OpenEDR is a sophisticated, free, open source endpoint detection and response solution. It provides analytic detection with Mitre ATT&CK visibility for event correlation and root cause analysis of adversarial threat activity and behaviors in real time. This world-class endpoint telemetry platform is available to all cyber-security professionals, and every sized organization, to defend against threat actors and cyber criminals.
Queries the OneCall API of OpenWeatherMap for a given location by latitude and longitude. This API returns information about the current weather, forecasts by minute, hour, and day, and national weather alerts.
A Huginn agent that writes events into InfluxDB in line protocol format (primarily). No docs right now but there are only two files and the Huginn on-line docs therein seem helpful.
The active_workflow_agent library helps you to write your own ActiveWorkflow agents in Ruby using ActiveWorkflow's remote agent API. “Remote” in this context means that agents run in separate processes from ActiveWorkflow itself. Communication between agents and ActiveWorkflow takes place via HTTP. Each agent is effectively an HTTP server which ActiveWorkflow connects to and interacts with via the remote agent API protocol.
Github: https://github.com/automaticmode/active_workflow_agent
The activeworkflow_agent library helps you to write your own ActiveWorkflow agents in Python using ActiveWorkflow's remote agent API. “Remote” in this context means that agents run in separate processes from ActiveWorkflow itself. Communication between agents and ActiveWorkflow takes place via HTTP. Each agent is effectively an HTTP service or microservice which ActiveWorkflow connects to and interacts with as long as it supports the remote agent API protocol.
Github: https://github.com/automaticmode/activeworkflow-agent-python
You can create your own agents by developing simple services that implement ActiveWorkflow's Remote Agent API. “Remote” in this context means that agents run in separate processes from ActiveWorkflow itself. Communication between agents and ActiveWorkflow takes place via HTTP. Each agent is effectively an HTTP service or microservice which ActiveWorkflow interacts with via an RPC protocol.
https://github.com/automaticmode/active_workflow/blob/master/lib/remote_agents.rb
https://github.com/automaticmode/active_workflow/blob/master/spec/features/use_remote_agent.rb
https://github.com/automaticmode/active_workflow/blob/master/spec/lib/remote_agents_spec.rb
Agent that monitors a flow of data and uses Isolation Forest techniques to identify statistical outliers and anomalies.
I'm using Flask to host some simple REST APIs for custom Huginn agents.
- expenses.py - Uses the Plaid API to get transactions and expenses filed.
- logger.py - The agents's logging service.
A distributed task queue written in Python that implements multiple message brokers and workers. Used to distribute tasks (discrete units of work or messages) to worker processes elsewhere on the host or the network. Generic communication protocol - there are protocol adapters for multiple other programming languages. Also implements webhooks.
Aims to be easy to use, with no required configuration files. This might be a case of "write your own simple daemon." Requires RabbitMQ or Redis as its message brokers.
https://docs.celeryproject.org/en/stable/
https://docs.celeryproject.org/en/latest/getting-started/first-steps-with-celery.html