Free guardrails and engineering discipline for AI coding: blocks leaked secrets (incl. AI-provider keys) before commit/push, destructive commands and insecure defaults; backs up before edits; verify-before-done; project continuity (init · checkpoint · resume); Coach mode; AI-use reports; self-testing doctor. Local only. English and Bahasa Malaysia.
9 skills · 1 command · hooks
Recover and update firmware on devices you own — CCTV cameras, recorders, routers, discontinued models: identify the hardware, back up the original flash first, recover through vendor tools, UART/U-Boot/TFTP or an SPI programmer, then secure the device.
1 skill
IoT and embedded work done safely: identify the board and its voltages before wiring, back up flash before flashing, MQTT and OTA with authentication and a way back, and proof on the real device before 'done'.
1 skill
PLC, industrial automation and robotics with safety first: safety functions stay with qualified engineers, programs are backed up from the PLC before editing, simulation before site, Modbus/OPC UA done right, commissioning evidence.
1 skill
Research with integrity: every reference verified to exist (DOI, Crossref, OpenAlex, PubMed, arXiv), accurate quotes and numbers, reproducible analysis, sound statistics, ethics and AI-use disclosure.
1 skill
Video, animation and media done reliably: ffmpeg recipes checked with ffprobe and playback, programmatic animation (Manim, Blender, Remotion), captions from speech, and clear rules on copyright and synthetic media.
1 skill