Test build for AI coaching agent

I am looking for a developer who can build a lightweight, hosted chat application that wraps an existing system prompt around the Anthropic Claude API and gives my testers a shareable URL. This is a testing and validation build. Please read the attached detailed spec.

Developer Brief: Purpose Coaching Agent — Test Build

I have developed and tested a new coaching agent's system prompt and am now ready to deploy a simple hosted version so a small group of nonprofit leaders can test it and give me feedback.

Requirement
A lightweight, hosted chat application that wraps our existing system prompt around the Anthropic Claude API and gives testers a shareable URL. This is a testing and validation build - I want the simplest thing that works well.

Scope - Build a chat web application with the following:
- Clean, mobile-responsive chat interface
- Calls the Anthropic API using Claude Sonnet 4.6 (model string: `claude-sonnet-4-6`)
- Our system prompt (provided separately) is loaded server-side and is not visible or accessible to end users
- Conversation history maintained within a session (no persistence across sessions required at this stage)
- A welcome message displayed when the user first opens a conversation
- Simple branding: our logo, our primary brand colour (#0F4C81 — Classic Blue), and the name "Purpose Coaching Agent"
- A brief footer or note indicating this is a prototype for testing purposes
- A simple passcode gate so that only people we give the passcode to can access the chat — nothing more than a single shared passcode stored as an environment variable
- The ability for us to update the system prompt without redeploying the application

Needs to deploy free or low-cost hosting with:
- A shareable public URL
- Our Anthropic API key stored securely (not in the codebase)
- No account creation or login required for testers beyond the passcode above

Not Needed
- User authentication or individual user accounts
- Conversation history saved to a database
- Admin dashboard or analytics
- Rate limiting or abuse prevention (access is controlled by passcode and we are sharing the link with fewer than 20 people)
- RAG, document retrieval, or knowledge base integration

Technology
Streamlit with Streamlit Community Cloud hosting or a Next.js or Flask app deployed to Vercel or Railway would be fine. If you want to use something else, let me know why and the benefits.

The system prompt should be stored in a configuration file or environment variable that I can edit directly (not hard-coded into application logic).

What I’ll Provide
- The full system prompt (a text document, approximately 3,000 words)
- Logo file (PNG, transparent background)
- Brand colour: #0F4C81 (Classic Blue)
- An Anthropic API key for development and testing
- Feedback and sign-off within 24 hours of deliverable

Deliverables
1. Source code in a private GitHub repository (transferred to our GitHub account on completion)
2. Deployed, working application with a live URL
3. Brief setup notes: how to update the system prompt, how to change the passcode, how to redeploy if needed, where the API key is stored

Timeline
Completed within five working days of commissioning with actual build of one to two days.

Budget
Please quote a fixed price.

Please include:
- A fixed-price quote and estimated delivery timeline
- One or two examples of similar work (a chat interface calling an LLM API, or any Anthropic/OpenAI API integration)
- The framework and hosting platform you would use, and why
- Confirmation that you're comfortable with the Anthropic API specifically