Skip to content

assignment.md

Latest commit

 

History

History
118 lines (81 loc) · 4.51 KB

assignment.md

File metadata and controls

118 lines (81 loc) · 4.51 KB

Week 5 — Agentic Development with Warp

Use the app in week5/ as your playground. This week mirrors the prior assignment but emphasizes the Warp agentic development environment and multi‑agent workflows.

Learn about Warp

Explore the Starter Application

Minimal full‑stack starter application.

  • FastAPI backend with SQLite (SQLAlchemy)
  • Static frontend (no Node toolchain needed)
  • Minimal tests (pytest)
  • Pre-commit (black + ruff)
  • Tasks to practice agent-driven workflows

Use this application as your playground to experiment with the Warp automations you build.

Structure

backend/                # FastAPI app
frontend/               # Static UI served by FastAPI
data/                   # SQLite DB + seed
docs/                   # TASKS for agent-driven workflows

Quickstart

  1. Activate your conda environment.
conda activate cs146s
  1. (Optional) Install pre-commit hooks
pre-commit install
  1. Run the app (from week5/ directory)
make run

  1. Open http://localhost:8000 for the frontend and http://localhost:8000/docs for the API docs.

  2. Play around with the starter application to get a feel for its current features and functionality.

Testing

Run the tests (from week5/ directory)

make test

Formatting/Linting

make format
make lint

Part I: Build Your Automation (Choose 2 or more)

Select tasks from week5/docs/TASKS.md to implement. Your implementation must leverage Warp in both of the following ways (more details below):

  • A) Use Warp Drive features — such as saved prompts, rules, or MCP servers.
  • (B) Incorporate multi-agent workflows within Warp.

Keep your changes focused on backend, frontend, logic, or tests inside week5/. For each selected task, note its difficulty level.

A) Warp Drive saved prompts, rules, MCP servers (REQUIRED: at least one)

Create one or more shareable Warp Drive prompts, rules, or MCP server integrations tailored to this repo. Examples:

  • Test runner with coverage and flaky‑test re‑run
  • Docs sync: generate/update docs/API.md from /openapi.json, list route deltas
  • Refactor harness: rename a module, update imports, run lint/tests
  • Release helper: bump versions, run checks, prepare a changelog snippet
  • Integrate the Git MCP server to have Warp interact with Git autonomously (creating branches, commits, PR notes, etc)

Tips: keep workflows focused, pass arguments, make them idempotent, and prefer headless/non‑interactive steps where possible.

B) Multi‑agent workflows in Warp (REQUIRED: at least one)

Run a multi‑agent session where separate agents in different Warp tabs handle independent tasks concurrently.

  • Perform multiple self-contained tasks from TASKS.md in separate Warp tabs using concurrent agents. Challenge: how many agents can you have working simultaneously?

Tips: git worktree may be helpful here to keep agents from clobbering over each other.

Part II: Put Your Automations to Work

Now that you’ve built 2+ automations, let's put them to use! In the writeup.md under section "How you used the automation (what pain point it resolves or accelerates)", describe how you leveraged each automation to improve some workflow.

Constraints and scope

Work strictly in week5/ (backend, frontend, logic, tests). Avoid changing other weeks unless the automation explicitly requires it and you document why.

Deliverables

  1. Two or more Warp automations, which may include:

    • Warp Drive workflows/rules (share links and/or exported definitions) and any helper scripts
    • Any supplemental prompts/playbooks used to coordinate multiple agents

  2. A write‑up writeup.md under week5/ that includes:

    • Design of each automation, including goals, inputs/outputs, steps
    • Before vs. after (i.e. manual workflow vs. automated workflow)
    • Autonomy levels used for each completed task (which code permissions, why, and how you supervised)
    • (if applicable) Multi‑agent notes: roles, coordination strategy, and concurrency wins/risks/failures
    • How you used the automation (what pain point it resolves or accelerates)

SUBMISSION INSTRUCTIONS

  1. Make sure you have all changes pushed to your remote repository for grading.
  2. Make sure you've added both brentju and febielin as collaborators on your assignment repository.
  3. Submit via Gradescope.