gptme-webui
gptme-webui copied to clipboard
gptme
Web UI for gptme, built with lovable.dev
The first command executed in the shell does not return any output and blocks forever. User has to type another message and then command can reexecute and work.
## Overview PR gptme-webui#95 has three categories of issues that need to be addressed: ## 1. Critical: Error Handling Architecture (Greptile's Finding) **Problem**: The error handling code in `WelcomeView.tsx` (lines...
Must be related to the placeholder conversation created before the conversation is created on the server, probably happened in https://github.com/gptme/gptme-webui/pull/89
Update Vite configuration to set the development server host to 0.0.0.0 and port to 5000, and allow all hosts. Add replit.md documentation detailing the project setup and configuration for Replit....
in line with modern design standards for AI chat apps
Interesting linter, not sure how many of the failing lints are false positives though.
Similar to the autocomplete in the CLI. Should be easy to do since we now have the workspace API.
A more meaningful message such as "User chose not to execute tool" would prevent the llm resuggesting execution of the same tool call.
Incase `gptme-webui` is launched before the server, it should try connecting to the default url periodically. Better UX
As I've seen and @0xbrayo [reported](https://discord.com/channels/1271539422017618012/1347573675095756970/1371414005268217977), sometimes the input box gets stuck in disabled with a "Waiting for chat session to be established..." placeholder, until refresh or such. I think...