This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle
What I Built
"Hey, could we also add online ordering and payment? It should be a small change, right?"
If you're a freelancer, you've probably encountered a request like this. What sounds like a quick addition can quietly turn into hours of extra work—and an awkward conversation about whether it was included in the original price.
ScopeShield helps freelancers compare new client requests against their original project agreements.
It analyzes the agreed scope, identifies what appears to be included, potentially additional, or in need of clarification, and highlights relevant agreement clauses. It can also help draft a professional response that explains the situation and opens a conversation about the extra work.
The goal isn't to automatically reject every new request. It's to help freelancers understand what changed, communicate boundaries with confidence, and make an informed decision before committing to work that may fall outside the original agreement.
Demo
The project was supposed to be a bakery website. Then came the message that changed the scope.
A freelancer has agreed to redesign The Daily Crumb's homepage, menu, and contact page. The agreement is clear: two rounds of revisions are included, but online ordering, shopping carts, and payment integration are not.
Then the client sends this:
"The new website looks great! Could we just add online ordering, a shopping cart, and payment functionality before launch? It would make things much easier for our customers. Since we're already working on the website, I thought these would be quick additions. Can we include them in the current project?"
It's a familiar freelancer dilemma. You want to keep the client happy, but you also don't want a "quick addition" to quietly become hours of unpaid work.
That's where ScopeShield comes in.
Instead of relying on memory or rereading the entire agreement, the freelancer can compare the new request against the original scope. ScopeShield flags the online ordering and payment features as potentially outside the agreement, highlights the relevant exclusions, and explains why the request needs a closer look.
But identifying scope creep is only half the problem. The other half is figuring out what to say without sounding difficult or confrontational.
ScopeShield can draft a polite response that acknowledges the client's idea, explains that these features weren't included in the original agreement, and opens a conversation about requirements, timeline, and estimate.
The freelancer doesn't have to say no to the opportunity. They can say yes to the conversation—without accidentally saying yes to the extra work for free.
Code
KrishangeeTayal
/
ScopeShield
A local AI-powered scope-creep detector for freelancers.
ScopeShield
Know What's Included. Before You Say Yes.
ScopeShield is a local AI-powered scope-creep detector for freelancers. It compares new client requests against an original project agreement, highlights potential scope changes, and helps draft polite, professional replies.
Features
- Analyze client requests against an agreed project scope.
- Identify requests that appear included, potentially additional, or need clarification.
- Highlight relevant agreement evidence and explain findings.
- Draft professional replies to clients.
- Keep a Scope Ledger of previous analyses.
Tech Stack
- Next.js
- TypeScript
- Tailwind CSS
- Gemma 3:4b
- Ollama for local AI inference
Prerequisites
Before running ScopeShield, install:
Installation
1. Clone the repository
git clone https://github.com/KrishangeeTayal/ScopeShield.git
cd ScopeShield
How I Built It
ScopeShield is built with Next.js, TypeScript, and Tailwind CSS.
The AI analysis runs through Gemma 3:4b using Ollama, with inference on the local machine rather than a hosted LLM API.
The application compares the project scope with a new client request and produces findings that include:
- Whether a request appears to be included, potentially additional, or needs clarification
- Relevant agreement evidence and clauses
- An explanation of why the request was flagged
- Whether human review is needed
I also built a reply-drafting feature. A freelancer can generate a response that acknowledges the client's request, refers to the agreed scope when relevant, and discusses next steps without assuming that every extra request is automatically billable.
A Scope Ledger preserves previous analyses so freelancers can refer back to earlier findings.
Why Does Open Innovation Matter?
For ScopeShield, running an open-weight model locally is central to the product.
Freelancers may be working with client contracts, project estimates, business requirements, and private conversations. Sending all of that information to an external AI API may not be desirable.
By using Gemma through Ollama, ScopeShield can analyze this information locally, keeping the workflow on the freelancer's machine rather than requiring a hosted inference service.
The local approach also avoids per-request API charges and gives me control over the model, prompts, and analysis workflow. I can experiment with different open-weight models and adapt the system to the language freelancers use in project agreements.
The trade-off is that local inference requires a machine capable of running the model, and model outputs still need human review. ScopeShield is a decision-support tool, not a substitute for reading an agreement or making a contractual determination.
My First Client
I showed ScopeShield to a potential user who could relate to the problem of freelancers being asked to do extra work without being paid fairly.
They liked how straightforward the tool was and how it helped draft a polite message to the client, making it easier to communicate boundaries without sounding confrontational.
The feedback reinforced what I wanted ScopeShield to do: help freelancers understand what's included in an agreement and respond confidently when a new request comes along.
Prize Categories
- Best Use of Gemma — ScopeShield uses Gemma 3:4b locally through Ollama to analyze project agreements, identify potentially out-of-scope client requests, and help freelancers draft professional responses.
ScopeShield is also submitted for the overall Hacktoberfest Open-Source AI Challenge: Week 2.
Top comments (4)
Do check out!
Great work Krishangee!!
Excellent 💗
Wonderful work!🩷