Semiconductor Processing Technician AI Expert Role - Transform Human Expert Knowledge into AI Agents

Finally Found a Tool to "Copy" Semiconductor Technician Experience into an AI!

Hello everyoneI'm your old frienda struggling engineer who deals with equipment all day in a semiconductor fabRecentlyour company wanted to go digital and asked us veteran technicians to write down our experienceIt gave me a headache – those process parameters and troubleshooting methods are too boring to documentand newcomers can’t really understand them anywayUntil I stumbled upon wonsukchoi's domain-experts project onlinewhich contains a semiconductor processing technician expert skill – that totally blew my mind!

What exactly is this skill? Simply putit packages a semiconductor technician's expertise into an AI AgentYou install it intoand can chat directly with a "virtual veteran technician." Whether it’s lithography alignmentetching rate tuningCMP slurry ratioor even how to trace common particle contaminationit explains everything clearlyThis is way better than the training system our company spent hundreds of thousands on!

How to Install: Two Stepsas Simple as Instant Noodles

HonestlyI was a bit worried the installation would be complicated – after allI’m a process engineernot a developerBut after tryingit’s really just two commands.

npm install -g skills# Install the skills CLI tool (if not already installed)
npx skills add [repository-url] --skill semiconductor-processing-technician

Wait for the download to finish and you'll have a semiconductor technician AIYou can also add author information with the command below for easier updates later:

npx skills add [repository-url] --skill semiconductor-processing-technician --author wonsukchoi

Then how to use it? Just run it inor start interactive mode with skills run semiconductor-processing-technicianThe first time I opened itI asked a tough question: "What to do if the photoresist edge bead is too thick after coating?" Guess what? It listed 4 possible causes and corresponding adjustment methodsincluding spin speedbaking temperatureexhaust flow… almost identical to what the shop floor veteran used to tell me!

Real-World Scenario: Training Newcomers with an AI

Our department recently got three internsBeforeI had to personally mentor themanswering endless basic questions – super annoyingNow I deployed this skill onto our internal AI platform and told the interns to ask it whenever they have questionsThe effect is surprisingly good!

• Process Parameter Queries: An intern asked"What RF power is typically used for dry etching?" The skill directly gave typical rangesdifferences across materialskey points to noteand even referenced industry standards.
• Fault Simulation: I asked an intern to describe a failure symptome.g"Scratches on wafer surface after CMP." The skill guided them step by step to analyze possible contamination sourcescheck pad conditionand evaluate slurry residue.
• Equipment Operation Guidance: The skill contains many equipment operation stepslike "How to calibrate the lithography alignment system." It breaks down each step clearlyincluding safety precautions.

What surprised me most was that it can simulate a "reverse questioning" mode: when the user asks an incomplete questionit asks back for more specific detailsjust like a real technician troubleshootingThis interaction is far better than rigid FAQs.

Behind the Scenes: A Brief Look at the Tech

This skill is essentially a structured expert system combined with a large language modelIt consists of two parts: a knowledge base (containing domain-specific semiconductor process knowledgerulesand examples) and a prompt template (to make the AI "act" like a technician)The authorwonsukchoiclaims to "convert all human experts into AI agents." I looked at the repository structure and indeed found other expert roles under the roles directorylike mechanical engineerpharmacistetcIt feels like a large knowledge base ecosystem.

For us userswe don’t really need to care about the underlying implementation – just know how to use itAnd since it's open sourceif you find the knowledge insufficientyou can modify the SKILL.md file to add your own factory-specific process parameters or equipment modelsI added the recipe for our factory's brand of etcherand now it's become our department's dedicated AI consultant.

Notes and Pitfall Avoidance Guide

After using it for a weekI stepped into a few pitfalls – sharing them so you can avoid the detours:

• Network Issues: The first installation got stuck for a long time due to slow access to the repositoryI recommend using a proxy or manually downloading the ZIP from the repository's main page and extracting it to the skills directory(Direct download link omitted as per policy.)
• Runtime Environment: Requires Node.js 14+ andIf you just want to testyou can use skills run in the command line without opening the GUI.
• Privacy Concerns: This skill does not connect to the internet by default; all knowledge is localBut if you modify SKILL.md and add sensitive process data…