is there a list of actual commands that I can look at to get an agent to trigger an oauth inline via lindychat so I can give it access to a new account? it pisses me off so much we cant trigger it inside the agent.
Hi Gabe, unfortunately there is no native way to manually configure the skill auth within an AI Agent state. However, if you do have the valid authentication, you can prompt the agent to use a specific account
Can this please be added
I genuinely don't understand why you couldn't pass an additional modal to the skills of an agent. You're already referencing it there in display why couldn't that also be clickable and have a modal there. That would also protect people from poor prompting God forbid they forget something as well and it uses or attempts to use the wrong account
I feel like this is really just a quality of life request that should really be frontlined.
If there's not a way to do this is there a way to have a separate workspace? I'm paying like $400 a month and I'm stuck with one workspace, and tons of different projects. I love the platform but this is the most frustrating point that I have had in my time using it, being able to share multiple authorizations and getting the OAuth to trigger correctly in live chat and also being stuck with one workspace with the inability to provide a clear cut separation of auth
And I genuinely apologize if I just come off as a b***** user, I absolutely love your platform and it is truly one of the best out there because it responds so well to semantic and natural language. It's just this matter of managing authorization for different agents that is my biggest issue
Hi Gabe F. , This exact request has come up before and I agree with you. I've actually pushed for this feature internally as well. Manual auth config in AI Agent states would be a massive QoL improvement, and you're right about it protecting against prompting errors too. Just passed your feedback along about both the auth configuration and workspace limitations. Obviously a lot of factors go into prioritizing the roadmap and we are pushing as fast as we can 🙂, but really appreciate you taking the time to lay this out so clearly - detailed use cases like yours help make the strongest case for these improvements. Thanks! 🙏
Ethan K. I just really dont want to go and learn another platform but at this point it makes it extremely hard to build agents for clients. let me explain how we do this. our clients comm through private slack channels on our slack. we add an app to the channel, so they can @xxxxxx the app, then in the flow we just name the trigger filter for the agent the same as that app, even though it doesnt pass anything through the app, it just makes it easier for them to understand how to use the agent. we collaborate with them inside of slack with their team and get results directly from the agent, live and they can use as they see fit inside of that channel. now you can understand why something like the authing issue not being super clear and easy to manage, causes a problem. dumping the tools into an agent skill set along with a prompt is 1000x faster than building an entire workflow and it gives them additional freedom to ask the agent to do certain things. its created an extremely sticky offer for us because we can add another agent to their lindy for other tasking they need. it allows us to build their agents for them, and them not need to learn a platform, and we can work with them and troubleshoot live. I have one lindy that has 4 different agents in it, that can handle almost ALL of my work flow and tasking. i have clients that need this, but we want to roll it into the rest of our stack because of how sticky our offer is becuase of it.