Ask · Plan · Agent
Ask for read-only repository understanding, Plan changes before mutation, and use Agent for reviewed edits, approved commands and verification.
TuxNest is a local-first AI engineering environment for VS Code that helps developers understand repositories, plan changes, review edits, execute approved commands and verify results. I created it, own the product, and lead its ongoing development.
Creator · Owner · Lead Developer
I drive TuxNest’s product direction and engineering: the VS Code extension, local/remote model connections, reviewable edits, execution approvals, diagnostics and verification workflows. The product website and source repository are linked above so visitors can verify the project and inspect the code directly.
The goal is not just model chat; it is a visible engineering workflow.
Ask for read-only repository understanding, Plan changes before mutation, and use Agent for reviewed edits, approved commands and verification.
Connect to local Ollama, an explicitly configured compatible endpoint, or an Ollama runtime hosted on your own remote GPU through SSH.
Inspect proposed multi-file diffs, check snapshots before applying changes, and use recorded recovery data to undo accepted changes when available.
Keep edits, commands, internet access and sensitive or destructive actions behind explicit policy and review.
Distinguish saved-file readback, real command results, tests and browser checks from model-generated statements.
Discover and manage models, check runtime health, run self-tests, and review diagnostics before scaling a task.
Local inference uses the runtime you configure. Remote SSH mode sends prompts and included context to your selected host; API mode sends them to your configured endpoint. TuxNest does not operate a hosted inference service.
Terminal commands use your existing account permissions. Reviewed proposals and approvals support safer workflows, but they are not an operating-system sandbox.
Read the official product overview ↗