AutoProber: AI-driven flying probe automation for hardware hacking
What AutoProber does
AutoProber is a hardware hacker's flying probe automation stack designed to give AI agents everything needed to go from detecting a new target on a plate to safely probing individual pins. The system handles target discovery, microscope mapping, safety-monitored CNC motion, probe review, and controlled pin probing.
Workflow and operation
The typical workflow involves telling the agent to ingest the project, connecting all hardware, having the agent confirm all parts are functioning, running homing and calibration, attaching the custom probe and microscope header, and then notifying the agent of a new target on the plate. The agent will find the target location, take individual frames while recording XYZ coordinates, note pads, pins, chips, and other features, stitch frames together, annotate the map, add probe targets to the web dashboard for approval or denial, probe approved targets, and report back.
Hardware control and safety
All hardware can be controlled through the web dashboard, Python scripts, or by the agent itself. The project treats hardware movement as a machine-control system rather than a normal web app. The safety model requires continuous monitoring of oscilloscope Channel 4 during any motion, with any Channel 4 trigger, ambiguous voltage, CNC alarm, or real X/Y/Z limit pin serving as a stop condition. Recovery motion is not automatic.
Repository structure
apps/- Operator-facing scripts and Flask dashboard entrypointautoprober/- Reusable Python package for CNC, scope, microscope, logging, safetydashboard/- Single-page web dashboarddocs/- Architecture, device references, operations, and safety guidancecad/- Printable STL files for the current custom toolheadconfig/- Example environment/configuration files
Hardware stack
The tested architecture uses:
- GRBL-compatible 3018-style CNC controller over USB serial
- USB microscope served by mjpg_streamer
- Siglent oscilloscope over LAN/SCPI for Channel 4 safety monitoring and Channel 1 measurement
- Optical endstop wired to an external 5V supply and oscilloscope Channel 4
- Optional network-controlled outlet for lab power control
Reference parts
The prototype uses specific parts including optical end stop, USB microscope, SainSmart Genmitsu 3018-PROVer V2, Matter Smart Power Strip, Siglent SDS1104X-E Oscilloscope, Dupont wires, pen spring or similar light compression spring, and 3D printer for printable toolhead parts.
License and status
The repository is a self-contained source-available release candidate under PolyForm Noncommercial 1.0.0 license with commercial contact available. It uses Python with dependency resolution via uv.lock.
📖 Read the full source: HN AI Agents
👀 See Also
Developer Builds LibraHQ App to Solve AI Agent Memory Problem
A developer created LibraHQ, a free notes app that serves as a shared memory layer between chatbots and coding agents. The app records important notes and decisions from chats and stores them for future sessions, addressing the problem of AI agents forgetting previously made decisions.
Superglue CLI: Let AI Agents Execute API Calls Without Pre-Built Tools
Superglue CLI provides a skill that teaches AI coding agents how to use its commands, handle authentication, build tools, and debug failures. Instead of creating pre-built tools for every API integration, agents can read API specs at runtime and plan multi-step calls.
cq: A Local-First Knowledge Sharing System for AI Coding Agents
Mozilla.ai's cq is an open-source tool that lets AI coding agents share 'knowledge units' about common gotchas via a local SQLite store, with optional team sharing through a Docker API. It installs as a Claude Code plugin or OpenCode MCP server.
claude-sessions: Terminal UI for Browsing Claude Code Transcripts
claude-sessions is an open-source terminal UI tool that scans local Claude Code transcript files, allowing developers to browse, search, and resume past sessions. Built with Claude Code itself, it features WASD navigation, keyword search, and one-click session resumption.