Hi Product Hunt — I’m the maker of Agenaxy. You have Ollama or LM Studio running. The next question is: how do you give that model a useful set of tools and a place to work? Agenaxy brings those pieces together in a Mac app. Add your model Connection in the UI, select files for a Chat, and ask for a deliverable. File tools, shell execution, and readable Activity are built in. Standard includes web search without a separate search API key or search-server setup. For sensitive tasks, switch to the separate Vault workspace. You explicitly authorize the model Connection. The Agent’s web tools are removed, and macOS blocks network access from Agent-run scripts. Approving an action cannot open that network boundary. An authorized remote model still receives the context assembled for its call. The gallery shows a local Qwen run that executed a prepared analysis script over three synthetic CSVs and saved a Markdown report plus an exceptions file. All three source hashes matched before and after. A separate network test shows Standard reaching a test receiver while Vault is denied. The scope is focused: this beta does not include browser control, schedules, or a broad connector catalog. You still need to run a compatible model endpoint or supply provider credentials. The beta is free for Mac. Remote-provider usage is billed by the provider. If your local model is already running, what would you want it to do with your files first?

Agenaxy

Your local model, ready for real file work

Your local model, ready for real file work