coreinsight-cli 0.1.3

Local-first AI performance profiler that mathematically verifies optimizations for Python, C++, and CUDA

CoreInsight CLI

CoreInsight is a local-first, hardware-aware AI performance profiler. It shifts performance engineering "left" by parsing your Python, C++, and CUDA code, identifying hardware bottlenecks (like CPU cache thrashing or CUDA warp divergence), and mathematically verifying AI-generated optimizations inside secure Docker sandboxes.

Prerequisites

• Python 3.9+
• Docker Desktop / Docker Engine (Must be running for the sandbox verification)
  • Install Docker: https://docs.docker.com/engine/install/
• Ollama (Optional, if using local models) or API keys for cloud models.
• Suggested: Setup OpenAI/Anthropic/Google API keys to use those models

Install

pip install coreinsight-cli

Usage

1. Build Locally: Clone this repository and install it in editable mode:

pip install -e .

2. Configure CoreInsight CLI: Set up your preferred AI provider (Ollama, local vLLM, OpenAI, Anthropic, or Gemini):

coreinsight configure

3. Build Global Context (Recommended for multiple files): Index your repository so the AI understands your custom structs, classes, and dependencies across files:

coreinsight index

4. Test on a file: Analyze a specific file. The CLI will extract hot loops, process them in parallel, verify optimizations in Docker, and output a live Markdown report.

coreinsight analyze <file_name>

5. Project-Wide Hotspot Scanning: Instead of guessing which files are slow, scan your entire repository. CoreInsight will use static AST analysis to rank the most complex, deeply-nested loops in your project.

coreinsight scan

Build the Project

Download build:

pip install build

Run the build command to generate wheel file:

python -m build --wheel

To build project elsewhere using wheel file:

pip install dist/coreinsight_cli-*.whl

Architecture Notes

CoreInsight runs 100% locally. Code is only transmitted to the AI provider you configure. If you use Ollama or a local server, your proprietary code never leaves your machine.