INTAKE_QUESTIONS.md

skills/cowork-project-architect/INTAKE_QUESTIONS.md

1. # Intake Question Bank
3. Reference for the intake phase. The SKILL.md tells Claude when to ask which batch. This file is the bank to draw from when adapting questions to specific project types.
5. ## Universal (always relevant)
7. - What's the project's job, in one sentence?
8. - Who uses it? (Solo / collaborators / client-facing)
9. - Mode? (Interactive / agentic / hybrid)
10. - Cadence? (Daily / weekly / project-bounded / on-demand)
11. - What does Claude produce?
12. - What data lives in the project?
14. ## Vault structure
16. - Any folder organization that should match an existing vault you have?
17. - Are there subdirectories that need their own scoped CLAUDE files? (Examples: a `clients/` folder with per-client rules, a `campaigns/` folder where each campaign has its own brief format, a `domains/` folder for multi-domain research)
18. - Tag taxonomy you want me to seed, or build it as we go?
19. - Do you use any specific Obsidian plugins (Dataview, Templater, Tasks) that the templates should account for?
21. ## Project-type specific
23. ### Research / investigation projects
24. - What's the research question or hypothesis?
25. - What sources are in scope? Out of scope?
26. - Is source-tier discipline required? (i.e., distinguishing primary, secondary, tertiary)
27. - What's the publication target: internal notes, public writing, or both?
28. - Are there sensitive entities (people, places) that need special handling?
29. - Should each entity / subject get its own subfolder with a scoped CLAUDE file?
31. ### Writing projects
32. - What's the voice: yours, ghostwritten, or somewhere between?
33. - What's the audience?
34. - What length and format are typical outputs?
35. - Do you want Claude to draft from scratch, edit your drafts, or both?
36. - Drafts vs. final outputs: separate folders or same folder with status tags?
38. ### Code projects
39. - What language(s) and frameworks?
40. - Solo or multi-contributor?
41. - Test coverage expectations?
42. - Production code or experimental?
43. - Is Claude writing, reviewing, or both?
44. - Will the vault hold the code itself, or just specs and notes?
46. ### Data / pipeline projects
47. - What's the data shape (volume, format, source)?
48. - Is processing one-shot or recurring?
49. - Where do outputs go?
50. - What's the failure mode if data is bad?
51. - Should working data and reference data live in different subfolders with their own rules?
53. ### Communication / outreach projects
54. - Who's the audience?
55. - What's the tone?
56. - Is each output reviewed before sending, or is Claude trusted to send?
57. - What contact data sources are involved?
58. - Per-client or per-campaign subfolders with scoped CLAUDE files?
60. ### Design / creative projects
61. - What deliverable formats?
62. - Brand or style constraints?
63. - Iteration vs. first-pass production?
64. - Reference materials or moodboards in the knowledge base?
65. - Do mood/inspiration files need their own subdirectory rules (e.g., "don't paraphrase, just link")?
67. ## Always ask before generating
69. - Anything to deliberately exclude from scope?
70. - Existing skills, templates, or projects this should plug into?
71. - Any sensitivity (privacy, copyright, confidentiality)?