I will build and stress test your claude code ai workflow with proof checks


About this gig
I build one real task of yours into a working, tested tool for Claude Code.
WHAT YOU GET
- Your task as a script you can run with one command
- Two proof checks made from your own examples
- A stress test: rough inputs (empty files, missing columns, 10,000 rows) tried before delivery
- A health report you open in your browser
- Save points you can go back to
- A guided walk, and a tool to add your next workflow
HOW IT WORKS
1. Message me your task in one sentence. I confirm it fits before you order.
2. Send two small made-up examples and the correct result for each.
3. You receive a ZIP and run it yourself. I never need your computer, logins or real data.
The script itself is plain Python with no AI calls. Claude Code is the AI you use to run, check, explain and change it.
GOOD FIT: one local task with an input file and a result you can check.
NOT A FIT: whole apps, websites, deployments, chatbots, or anything needing logins or online services.
YOU NEED: python3 3.10+, git, your own Claude Code. Tested on Ubuntu Linux; written for macOS and Windows-WSL, not yet tested there. Not a security audit.
Independent service. Not affiliated with or endorsed by Anthropic.
Get to know Lawrence K
Founder, Visual Story Factory
- FromUnited States
- Member sinceFeb 2014
Languages
English
My Portfolio
FAQ
Does the tool itself use AI when it runs?
No. The script is plain Python that gives the same result every time, with no AI calls, logins or online services. Claude Code is the AI you use to run it, check it, explain it and change it. Your own scripts or a scheduler can also run it with one command.
Will it work on my computer?
You need python3 3.10 or newer, git, and your own Claude Code. Tested on Ubuntu Linux. Written to also run on macOS and on Windows with WSL, not yet tested there. Plain Windows without WSL is not supported. Message me the output of: python3 --version; git --version; claude --version; uname -sr
Will it fit my task?
It fits if your task has an input (a file or folder) and a result you can check. Example: a sales export in, a summary with totals out. It does not fit whole apps, websites, deployments, or anything that needs logins or online services.
Do you need access to my computer or my data?
No. You run everything yourself. Send two small made-up examples; never real customer data, passwords or keys.
What exactly do I receive?
A ZIP: your task as a runnable script set up for Claude Code, two proof checks built from your examples, a stress-test report, a health report page, save points, a guided walk, and a tool to add more workflows later.
What does a revision cover?
A wrong result on the examples we agreed. Basic includes 1 revision, Standard 2 and Premium 3. New examples or new features are a new order.
How fast will I know if it fits?
Message me before ordering. You get a clear yes, a narrower version, or an honest no in one reply.
What do I need installed?
python3 3.10 or newer, git, and Claude Code installed and signed in with your own account. Nothing else to install: the kit uses only Python's standard library. Linux is tested; macOS and Windows with WSL are written for but not yet tested. Plain Windows without WSL is not supported.

