Skip to content
View vidoluco's full-sized avatar
💻
Claude Maxxing
💻
Claude Maxxing
9 followers · 9 following

Achievements

Achievement: Pull SharkAchievement: YOLO

Achievements

Achievement: Pull SharkAchievement: YOLO

Block or report vidoluco

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
vidoluco/README.md

Ludovico Cesaro. Agents that do real work on a Mac. Local first, output on disk.

AI Engineer. Agents on Google Cloud at work, local-first tooling for macOS here. I use Claude Code, Copilot, Cursor and Antigravity on the same codebases with one shared set of rules, so most of what I publish comes out of that setup: tools that capture, transcribe or guard data on the machine instead of sending it away.

Setup

Four coding CLIs read one private knowledge base on a Mac and write notes, transcripts and local commits to disk. git push is blocked by a shim.

Four CLIs, one Mac, one private knowledge base they all read. Unattended runs (loops, headless sessions, subagents) commit locally and stop there: a git and gh shim ahead of PATH exits 1 on anything that writes to a remote. Push and PR are manual.

lightweight-rec

Option R, then capture at 1 fps with the mic, Whisper on the machine, an optional pass through the coding CLI named in RECORD_AI for title, tags and summary, and a Markdown note in your vault.

Option+R. Screen and microphone recorded at 1 fps, about 110 MB an hour, transcribed with Whisper on the machine and filed as a Markdown note in the vault you already use. The capture never leaves the Mac. The title, tags and summary do, through whichever coding CLI you already have signed in, Claude Code, Cursor or Copilot (RECORD_AI=copilot), unless you set RECORD_AI=0.

git clone https://github.com/vidoluco/lightweight-rec.git && cd lightweight-rec && ./install.sh

toyota-telemetry

Every trip drawn on 3D terrain, harsh braking and acceleration marked, the places they repeat ranked beside them.

Your Toyota already records every trip, the electric share and every time it decided you braked too hard. The app shows one trip at a time and forgets the rest. This syncs the lot into SQLite on your Mac and answers what the app cannot: which junction you get wrong nineteen trips in a row, what leaving at 07:15 instead of 08:00 saves, what each place costs in fuel. No dongle, no subscription, no map keys, and a synthetic demo so it runs before you connect a car.

git clone https://github.com/vidoluco/toyota-telemetry.git && cd toyota-telemetry && uv run toyota demo

Smaller things
query-sanitizer-mcp MCP middleware. A local model redacts sensitive data from a prompt before it reaches an external LLM. Python
gfn-overlay macOS overlay that captures the screen and gets gameplay tips from a local vision model (LM Studio). Python
xbox-llm-controller A local LLM drives an Xbox controller on macOS. Python
Discord-mirror Mirrors a Discord channel to another one live, no elevated rights needed. JavaScript
TradingView-Indicators Pine Script indicators and strategies I actually use. Pine

Pinned Loading

  1. gfn-overlay gfn-overlay Public

    AI gaming overlay for macOS - captures screen frames and provides real-time gameplay tips via a local vision LLM (LM Studio)

    Python 1

  2. query-sanitizer-mcp query-sanitizer-mcp Public

    MCP middleware that redacts sensitive data via a local model before prompts reach external LLMs

    Python 1

  3. xbox-llm-controller xbox-llm-controller Public

    AI-powered Xbox controller that queries a local LLM to autonomously play games on macOS

    Python 1

  4. free-local-rag free-local-rag Public template

    [FULLY FREE] RAG chatbot using DeepSeek API and FAISS.

    Python

  5. modelcontextprotocol/servers modelcontextprotocol/servers Public

    Model Context Protocol Servers

    TypeScript 90.2k 11.6k