Email header parser limits
A teammate wants an AI agent to generate a regex-based parser to extract structured fields (sender, recipients, subject, attachments) from raw RFC 5322 email headers in Python. When would you stop the agent and not use it for this, and what failure modes make regex-from-AI the wrong approach here?
Implement
unfold_email_headers(header_lines: list[str]) → list[str]Examples
in
[["From: alice@example.com","Subject: Quarterly report"," and next steps","To: bob@example.com"]]out["From: alice@example.com","Subject: Quarterly report and next steps","To: bob@example.com"]in
[["Subject: Hi","","Subject: this line is body text"]]out["Subject: Hi"]in
[["To: dev@example.com,","\tops@example.com,"," qa@example.com"]]out["To: dev@example.com, ops@example.com, qa@example.com"]What a strong answer looks like
Treat the AI’s output as a draft to verify, not an answer to trust. Name the specific flaw and the input that triggers it, say how you’d catch it (tests, edge cases, reading critically), and how you’d re-prompt or decompose to get it right.
0:00 of about 16 min
Vibe & agentic: describe the solution in plain language (or narrate it) and the coach grades your approach.
Which questions mattered is sealed until you submit. Telling you now would just be handing over the edge cases.
Run or narrate your approach, then ask the coach.