Practical overview
When it fits
Use Sidekick when a conversation needs local document context, web research or local GGUF inference on a Mac. Select its API option when additional OpenAI-compatible remote models are needed.
Who it is for
- Mac users exploring local language-model chat
- Researchers and students working with documents and web sources
- Users who want a local-first AI workflow with optional remote model access
- Developers testing GGUF models or building agent-like workflows on macOS
Documented uses
- Chat with a local language model offline
- Ask questions about files, folders and websites
- Organize resources into configurable areas of interest
- Conduct multi-step web research and produce a report
- Use function calling for calculations or actions
Requirements & limits
The repository is public, marked MIT-licensed and supports OpenAI-compatible APIs. The documented local workflow is for macOS desktop use; no pricing or account requirements are specified.
- macOS is the documented desktop platform.
- The built-in llama.cpp engine supports modern GGUF local models.
- Files, folders and websites can be configured as conversational resources.
- OpenAI-compatible APIs add remote model choices but are distinct from offline local inference.
- Deep Research is documented to read 50–80 webpages before producing a report.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Set up a local GGUF chat workflow
Configure Sidekick to use a local model for conversational inference on macOS.
Before you start
- A Mac running macOS.
- A supported modern GGUF local model.
- Sidekick from its official repository.
Steps
Install and open Sidekick on macOS.
Use its built-in llama.cpp inference engine with a supported GGUF model.
Start a conversation with the local model.
Add files or folders as context, or drag files into the input field.
Expected outcome
A local-model conversation can use selected files and folders as supporting context.
Guide limitations
- The documented material does not specify macOS version requirements, model download steps or model performance.
- This guide covers the local workflow; OpenAI-compatible APIs are a separate option.
Configure an OpenAI-compatible model option
Add a remote model route alongside Sidekick’s local-first workflow.
Before you start
- An OpenAI-compatible API service.
- An API key for that service.
Steps
Open Sidekick’s provider settings.
Select a supported provider preset or another OpenAI-compatible API option.
Enter the API key.
Use the configured model in chat.
Expected outcome
Sidekick can use an OpenAI-compatible remote model in addition to local models.
Guide limitations
- The evidence does not identify provider-specific deployment requirements, pricing or compatibility details beyond OpenAI-compatible APIs.
- Using a remote model differs from the documented offline local-inference workflow.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 1 October 2026
- Original website title
- johnbean393/Sidekick · Official homepage title · retained in its source language