Local AI Toolkit

Lower costs. Faster results.
Project guide · Interactive website · Wiki · Support
Reduce paid cloud context and get to relevant code sooner. Local AI searches an existing repository index by meaning or symbol, and uses local models to inspect files, logs and images. Your agent receives concise findings and source locations instead of unnecessary source text.
Savings depend on the context actually needed for the request, including tool overhead. Response time depends on the task, index, queue and hardware; local inference is not automatically faster for every task.
A subscription to any cloud agent is optional: the same tools answer from the command line, if you would rather work on your own.
Contents
- Two ways to use it
- What a cloud agent gets
- What the command line gets
- Code search is not text search
- With a cloud agent: what stops costing tokens
- From the command line: not tokens, but time
- How it works
- Dependencies
- Installing
- Using it with a cloud agent
- Using it from the command line
- What language it answers in
- Updating and removing
- Impact evidence: C# pilot
- Documentation packages
- Building and testing
- Licence