Reporting is mostly the unglamorous work around the writing — research questions, story structure, interview prep, and relentless fact-checking. This pack gives you tested prompt templates for those stages, each with fill-in-the-blank variables and accuracy guardrails so the AI helps you think, not invent.
How it works
Each prompt names your story type, beat, and topic, and asks the model for research
help, structure, or self-editing — never for facts to be taken on trust. The bracketed variables —
[story type], [beat], [word count], [publication style] — shape the output to your
assignment. Filter by category or search, replace the brackets, copy, and paste into your AI tool.
Why accuracy guardrails matter
Language models confidently fabricate facts, dates, statistics, and even quotes. These prompts are written so the AI surfaces what to verify rather than asserting truth. Use it to generate research questions, outline structure, prep interviews, build fact-check checklists, and tighten prose — then confirm every checkable claim against a primary source before it enters a draft.
Tips for better output
- Treat every output as unverified. Especially names, numbers, dates, and anything that reads like a quote — the model can invent all of them.
- Use it before reporting, not instead of it. The angle, question, and outline prompts are strongest at the planning stage; the reporting still has to be yours.
- Paste real transcripts for quote pulls. Then check the model returned them verbatim — never let a paraphrase masquerade as a direct quote.
- Follow your outlet’s AI policy. Disclose use where required, and never pass AI-generated content off as original reporting.