Miranda's Project Guide

Project: PlainText Pal Category: Web App (Flask) + Anthropic API Last updated: April 14

Note: This guide is created based on the latest state of your project repository + any notes from our discussion. It may not always reflect the most up-to-date information.

Where You Are

Your spec is in great shape: "PlainText Pal" has a clear name, description, focused MVP features, and good inspiration references. You know what you want to build -- a website where users paste text and get readability suggestions powered by Claude.

What's missing is the scaffolding and your journal entry. Your repo currently has just the template files. The goal between now and Thursday is to get a running Flask app with a basic round-trip: paste text → get suggestions → display them.

What We Talked About

We mapped out the architecture:

  • Flask app -- A web form where users paste their writing
  • pal.py -- Your core module that talks to Claude. The key function:
    • def get_suggestions(text) → returns a dictionary with:
      • intent: the inferred intent of the writing (e.g. "professional", "academic")
      • suggestions: a list of strings with readability suggestions
  • Claude API -- Use Claude Sonnet via the Anthropic Python SDK

Next Steps (Before Thursday)

  1. Scaffold your project. Use this prompt with your agent:

    Read my project.spec.md and the Flask setup guide at
https://csc-121.path.app/unit-3/resources/flask-setup.guide.llm.md
Set up my project: initialize uv, install Flask and anthropic,
create the basic file structure, and build a minimal starting
point I can run.

  2. Create pal.py. Start with a simple version of get_suggestions(). Even a hardcoded response is fine as a first step -- you can wire up Claude after the Flask app works.

    Help me create a pal.py module with a function called
get_suggestions(text) that takes a string and returns a dictionary
with "intent" and "suggestions" keys. Start with a version that
calls the Anthropic API using Claude Sonnet. Here are the SDK docs:
https://platform.claude.com/docs/en/api/sdks/python

  3. Wire it together. Build one Flask route with a text box and a submit button. When the user submits, call get_suggestions() and display the results.

  4. Update your journal. Fill in the Checkpoint 1 section of project.journal.md -- what you did, what's next, and any blockers.

  5. Commit and push.

Checkpoint 1 Readiness

By Thursday April 16th, you need:

Helpful Resources