stableagents / stableagents

Stable, Semi-Autonomous, Reliable and Steerable LLM Agents for production use cases.

Stable Agents

Early access to the SLAM-1 model is only available upon request with a paid contract
email:[email protected]

Install Locally

First get an API Token:

https://platform.stableagents.com

Then install:

pip install stableagents

Agents are growing in popularity and while they do it's important to have a clear and comprehensive list of what they can and more important currently can't do. This repo exists to stay up to date with the latest advancements and use cases in AI Agents.

Contributions to this list are welcome. Before submitting your suggestions, please review the Contribution Guidelines to ensure your entries meet the criteria. Add links through pull requests or create an issue to start a discussion.

Future

In the future you will be able to hire and employee a Stable Agent for a wide range of tasks. Similar to how companies now use retool to cut time on internal tools, you will use Stable Agents to cut down on the time it takes to complete task without unncessarily hiring. The future billion/trillion dollar companies will be small robust teams, not large organizations

Organizations will be interested in powering everything with an LLM and to make this possible we will allow organizations that apply to use Stable Agents to generate revenue. Instead of charging a monthly fee or a yearly contract, we will put a cap on the amount of revenue/value a Stable Agent can add to an organization before revenue share is implemented

If you are one of these companies interested in revenue sharing, consider joining our [Discord](https://discord.gg/STrYnEDu9U) and type the name of your company in channel #revenue-generating-agnets

Stable Agents

Stable Agents are agents which are highly capable at complex reasoning. Eventually Stable Agents will be able to complete tasks with little to no supervision. Stable Agents will also be able to add LLM capabilites any non-intelligent software

Media Based Agents

• Autoclick "Skip Intro" on a browser based streaming service without having to click it

Productivity Agents

• Create a routine for tasks to be completed autoregressively

• Assign a task you would normally assign or hire for on a platform like UPWORK. We rate the complexity of the task before initiation to set proper expectations, Stable Agents is evovling and can't complete all tasks just yet

Coding Agents

• Init Stable Agents in your terminal to avoid dealing with boilerplate code [ 🚧 Not yet available]

---

Existing Agents

• Mind2Web Generalist Web Browsing Agent
• HyperWrite Personal Assistant
• Agent Ops
• Agent Hub
• Super Agent
• Lutra AI
• Nexus GPT
• AutoGen
• Meta GPT

Agent Capabilites (w/o SA)

• [?] Purchase and then summarize a book from Amazon.com
• [?] Donate to a twitch streamer
• [?] Updating an expired discord link for my webapp, copy and pasting the new link to my codebase on github and redeploy my webapp
• Shop for house and accurately find one baesd on my description
• Create a Spotify Playlist
• [?] Opening Google Docs and drafting that CBER/CDRH jurisdictional clarification letter I’ve been dragging my feet on sending to FDA.
• Reading the fact that there are a bunch of job offers to go out off of my Asana inbox, and opening tabs with the DocuSign offers I need to sign.
• Generating a pull request to fix some bugs or maybe pick off an Asana task in one of the web app projects See Codegen .
• Order on Doordash
• Send a Tweet
• Open a Youtube Video
• Run Payroll
• Adjust a time on my calendar and send a "I'm running late" email

License

If you have any questions regarding the license or using the repo: please email [email protected]. The mission is to empower anyone to use Stable Agents, while preventing anyone from making modifications or monetizing the repo for negative gain.