I freelance in Python. Every project starts the same way: open Cursor, paste my "context-setting" prompt, paste the "scope review" prompt, paste the "what could go wrong" prompt. Then I do it again on the next project.
I finally put all 118 of my most-used prompts into one PDF.
What's in it
- Bug root-cause (paste the traceback, get a hypothesis + minimal repro)
- Time estimates (turns "I think this is 2 days" into "this is 2 days, here are the 3 risks")
- .cursorrules author (writes your repo's rules file from a 3-sentence brief)
- Security audit (finds OWASP top 10 issues, ranks them, suggests fixes)
- Dockerfile generator (multi-stage, distroless, ~80MB final image)
- README writer (because nobody does it until they have to onboard someone)
- Test scaffolding (pytest + factory_boy, runs on first try)
- API client (typed Python client from OpenAPI spec)
- Migration script (Alembic, with rollback)
- Refactor planner (breaks a 2000-line file into 6 modules with no behavior change)
- Client comms (status update, scope creep, "I need more time" without sounding like a tool)
The "review before you code" prompt
If you only use one prompt from the kit, use this one:
Act as a senior Python architect. Review this spec and identify the 3 most likely failure points before we start. Assume we have 3 days, not 3 weeks. Be specific. Don't summarize the spec back to me — I wrote it.
This catches scope creep before it costs a day.
Free sampler (10 prompts): https://gobvan.gumroad.com/l/skzza
Full kit ($19): https://gobvan.gumroad.com/l/wypos
Top comments (0)