1 min lesson
Timed debugging drill
Show why this lesson detail matters: "The debugging screen is where support candidates and engineering-grade candidates separate."
Step 1 of 3
The debugging screen is where support candidates and engineering-grade candidates separate. They hand you a gnarly Cursor bug and watch whether you investigate like a scientist or thrash like someone refreshing and hoping.
Set a hard 30-minute clock. Pick a scenario with the texture of a real Cursor escalation, then narrate the whole arc out loud or in a doc you can replay: context, reproduce, isolate, evidence, hypothesis. The deliverable at the end is a clean report an engineer could act on without asking you a follow-up.
Learn more
Full explanation
The 30-Minute Investigation Arc
Interactive diagram. Step through it with the Next and Previous controls below, or Tab to a region to read its detail.
The order is fixed. Step through each phase to see what it produces under the clock.
Learn more
Full explanation
A sample scenario to run on yourself
A sample scenario to run on yourselfPick one and start the clock
TICKET #4471 - “Agent mode keeps deleting my imports” Reporter: paid team user, macOS, Cursor latest “Since this week, when I ask Agent to add a function, it sometimes strips unrelated import lines at the top of the file. Not every time. Happens more in big files. Started ~Tuesday.” Attached: one screen recording, no logs, no repro steps.
Localize: client vs server vs modelThe discriminating questions
- Layer
- Client (editor/extension)
- Evidence you'd want
- Repro offline, console errors, version bisect
- What confirms it
- Bug vanishes on an older build → client regression
- Layer
- Server (backend/request)
- Evidence you'd want
- HAR trace, request/response IDs, timestamps
- What confirms it
- Bad response returned for a clean request → server-side
- Layer
- Model (LLM behavior)
- Evidence you'd want
- Same prompt across models, temperature, diff size
- What confirms it
- Only one model produces it, others don't → model issue
| Layer | Evidence you'd want | What confirms it |
|---|---|---|
| Client (editor/extension) | Repro offline, console errors, version bisect | Bug vanishes on an older build → client regression |
| Server (backend/request) | HAR trace, request/response IDs, timestamps | Bad response returned for a clean request → server-side |
| Model (LLM behavior) | Same prompt across models, temperature, diff size | Only one model produces it, others don't → model issue |
For the imports bug, oversized diffs that drop content on big files smell like model + context truncation. Prove it by shrinking the file.
Learn more
Full explanation
The deliverable
The deliverableEngineering-ready report
TITLE: Agent mode drops unrelated imports on large files (>~600 LOC)
SEVERITY: High - silent data loss in user's source
REPRO (3/5 runs):
1. Open attached 700-LOC file
2. Agent: “add a debounce helper”
3. Observe: top import block truncated in returned diff
SCOPE: Reproduces on model X, not model Y. Started build 0.xx.
EVIDENCE: bad-diff.txt, request-id abc123, console.log attached
HYPOTHESIS: context truncation on large files drops the header region
before the edit is composed. Confidence: medium-high.
NEXT: confirm by capping file size; ask backend for truncation logs.Score yourself honestly on one thing: did you change one variable at a time or did you swap the model and the file size and the prompt all at once and then can't say which mattered? Every place you guessed instead of testing is your debugging gap. Circle it.
AI tools are usually allowed in this screen. Use Cursor to draft a repro script or summarize logs, then say out loud what you're verifying. The screen rewards driving an agent with judgment and penalizes pasting its output unread, so narrate the line where you'd reject a bad suggestion.