alisorcorp

alisorcorp / ask-local

Public

Delegate grunt work from Claude Code to a local LLM via LM Studio. File contents stay on your machine; only the final answer enters your Claude session.

18
3
100% credibility
Found Apr 20, 2026 at 18 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

ask-local enables delegating file reading, directory listing, and pattern searching tasks from cloud AI coding sessions to a local language model in LM Studio to avoid consuming cloud tokens.

How It Works

1
💡 Discover the code helper

You hear about ask-local, a clever way to let a local brain handle tedious file-checking jobs for your online AI coding buddy, saving precious thinking time.

2
📥 Grab and set up easily

Download the files and run the one-click setup to place the helper tools where your AI buddy can find them.

3
🧠 Wake up your local thinker

Start the free local AI app on your computer and load a smart model so it's ready to explore files.

4
💻 Chat with your AI about code

Open a conversation with your online AI coding assistant and point it to your project folder.

5
🔍 Ask local helper to dig in

Type a simple slash command like 'find all notes to fix' or 'list key parts of this folder' — your local brain jumps in to read and search without slowing your main chat.

6
📋 See results stream back

Watch as lists, summaries, and findings appear quickly, with notes on what it checked and how much it used.

🎉 Work smarter, spend less

You now analyze big projects, hunt bugs, or inventory code effortlessly, keeping your online AI fresh for the big ideas.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 18 to 18 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is ask-local?

ask-local delegates repetitive code tasks from Claude Code sessions to a local LLM via LM Studio, keeping file contents on your machine while only sending the final answer to Claude. You get a /ask-local slash command for inventorying repos, grepping TODOs or env vars, triaging logs, or extracting patterns—plus CLI tools for direct use. Python-based, it works with tool-calling models like Qwen, with budgets to cap reads and streaming for quick feedback.

Why is it gaining traction?

It crushes token waste in Claude (up to 30x savings on audits per benchmarks), unlike delegate github copilot or cloud-only agents, by caching reads, offering free grep/list_dir, and forcing synthesis before spirals. Devs dig the privacy (no code leaves your box), precise flags like --read-budget 15, and token footers—beats manual prompting or delegate ai github hacks. Hooks like one-shot installs and spot-check rules make grunt work like ask local inventories fly.

Who should use this?

Full-stack devs refactoring large repos, SREs classifying error logs, or security folks hunting secrets without cloud leaks. Ideal for Claude Code users hitting context limits on multi-file triage, like listing API routes or spotting N+1 queries. Skip if you lack LM Studio hardware for 30B models.

Verdict

Solid early pick for Claude power-users (18 stars, 1.0% credibility)—docs shine, but verify outputs as local models lag on nuance. Install if token compaction kills your flow; it'll extend sessions without delegate to agent github copilot drama.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.