EC1B1 Python Coding SupportFollow-up resources for LSE BSc Economics
← Back to all posts

20 Feb 2026

Session 5 Summary: Mock Coursework Workflow and AI Project Setup

Final support session (5 of 5) on mock coursework execution, AGENTS.md setup, reusable skills, and careful verification of AI-assisted Python work.

Session overview

This was Session 5 of 5 in the programme (Fri 20 Feb, 09:00 GMT). The session showed the full mock coursework workflow from start to finish: download materials, structure a project, create an AGENTS.md anchor file, define reusable skills, and use AI tools in a controlled way for Python replication work.

Key topics covered

1. Mock coursework page and materials

  • Students were asked to use the new Mock Coursework page and click Get started.
  • Core downloads reviewed in class:
    • coursework brief
    • dataset zip file
    • Python requirements file
    • AI replication guide
  • Recommendation: keep all files inside one project repository from the beginning.

2. Coursework context and deliverables

  • The practice project replicates parts of Palma (2022) using provided data files.
  • Submission components discussed:
    • runnable Jupyter notebook
    • PDF answers for sections 5 and 6
    • short video presentation
    • signed cover sheet
  • Group work and timeline reminders were given alongside the mini assessment deadline.

3. Read first, code second

  • The coursework brief should be read fully before writing code.
  • The data README should also be read first to understand variable definitions and documentation.
  • The session emphasized this as an exam-style discipline: read all instructions before starting.

4. Skills and sub-agents in practice

  • The workflow distinction was clarified:
    • Skills for repeatable, guided workflows (for example, data cleaning).
    • Sub-agents for more autonomous, specialized tasks.
  • A practical data-cleaning skill example was demonstrated:
    • run diagnostics
    • produce a report
    • propose fixes
    • request approval before executing changes

5. Why AGENTS.md matters

  • The session introduced AGENTS.md as the project-wide instruction file that AI tools read before tasks.
  • Key recommendation: keep it concise and precise.
  • Include rules and standards, not a long task dump.
  • Important example rule highlighted in the session: never overwrite raw CSV files.

6. AI usage and responsibility

  • Students can use AI for coding support, but remain responsible for correctness.
  • Generated outputs must be checked for reasonableness and economic plausibility.
  • Final reminder from the session: you are an economist first; AI is an assistant, not a replacement for judgment.

Recommended workflow from this session

  1. Download all coursework files into one repository.
  2. Read the brief and data documentation fully.
  3. Create a short AGENTS.md with project rules and coding standards.
  4. Define one or two high-value skills for repeated tasks.
  5. Execute in small chunks and review every output.
  6. Approve data fixes deliberately; do not auto-accept suggestions.
  7. Keep raw data untouched and version your work clearly.

Reminders for students

  • Never overwrite original raw CSV files.
  • Always read AI output before accepting it.
  • Check economic interpretation, not only code execution.
  • Practice with the provided notebooks and tools; repetition builds speed and confidence.

What's next

This was the final live support session in the series. Use the mock coursework resources, recordings, and session posts as your reference set while preparing for the full coursework workflow.

Related resources