Agent Cloud is like having your own GPT builder with a bunch extra goodies. The GUI features 1) RAG pipeline which can natively embed 260+ datasources 2) Create Conversational apps (like GPTs) 3) Create Multi Agent process automation apps (crewai) 4) Tools 5) Teams+user permissions. Get started fast with Docker and our install.sh
Is your feature request related to a problem? Please describe.
If a user wants to define a custom tool using custom Python, this can pose a serious security risk as arbitrary and possibly malicious code must be executed on the platform in order to use this tool.
Describe the solution you'd like
We would like to leverage Google Cloud functions to execute custom code tools that are defined by users. this will minimise the security risk that arbitrary code can pose.
[x] show error logs in the agent chat response when function fails
[x] show error reason for failed deployment inside the function tool editor
[x] send a notification over socket and use it to trigger a refresh of function states automatically
[x] limit tool selection dropdowns to disable/hide pending functions
[x] only make functions redeploy when something that affects the function changes (anything except "name" and/or "description")
[x] Update delete endpoint to delete function
[x] add "states" to tools (pending, ready, error) and dont wait for deployment to finish after saving
[x] render example code to the right of where the user is defining their custom code
[x] let users know of the required patter they must follow when defining an environment variable
[x] Backend check to prevent selecting a tool for a task with a state other than "ready"
Add versioning
[x] sidebar with revision history
[x] add functionrevisions collection with code, deployment date, tags/notes, etc
[x] allow selecting revision and show code to redeploy
[x] change functions screen to tabbed view like gcp functions
[x] Put revisions in their own tab
[ ] Separate view for create vs edit tool with stepper on create page, edit page is the current view
Is your feature request related to a problem? Please describe. If a user wants to define a custom tool using custom Python, this can pose a serious security risk as arbitrary and possibly malicious code must be executed on the platform in order to use this tool.
Describe the solution you'd like We would like to leverage Google Cloud functions to execute custom code tools that are defined by users. this will minimise the security risk that arbitrary code can pose.