Ship30for30: Turn 1 LinkedIn Topic into 100 Post Ideas (Using Claude)
Why this is in the vault
A complete, copy-paste-ready multi-step Claude prompt for generating 70+ LinkedIn content ideas from a single topic input. The email explains the prompt engineering logic behind each section — role assignment, vocabulary scaffolding, sequential output control — making it useful both as a working tool and as a reference for constructing similar chained prompts.
The Framework: Topic → Subtopics → Proven Approaches
The "Endless Idea Generator" system has three layers:
- Topic — one phrase, 5 words or fewer (more specific = better output; "Running your first 5k as a single mother of 3 who hasn't run in 5 years" beats "Running")
- Subtopics — 7 action-verb-led outcomes within that topic (Claude generates these; verbs from a 40-item list baked into the prompt)
- Proven Approaches — 10 quantifiable questions per subtopic, drawn from 30 question types (Tips, Steps, Mistakes, Habits, Frameworks, etc.), each requiring a specific count of items
Math: 7 subtopics × 10 questions = 70 content seeds. Answer one per day = 70+ days of material.
Prompt Construction Principles (the meta-lesson)
The newsletter walks through 4 parts of the prompt design — each principle is transferable:
- Part 1 — Frame and define: Assign Claude a named role; define key vocabulary before using it; use explicit MUST rules for format constraints
- Part 2 — Teach by example: Include a positive example AND a "what to avoid" example; bake in a verb list so Claude draws from a constrained set
- Part 3 — Enumerate the output types: Provide the complete list of 30 proven approach question types with a fixed count per type — prevents open-ended drift
- Part 4 — Sequential output control: Issue multi-step instructions with a pause-point ("confirm before proceeding") to stop Claude from jumping ahead
Content Flywheel
Answer → tweet → atomic essay → YouTube short → thread → newsletter → long-form YouTube. Each answer to a proven-approach question can branch into multiple formats without additional ideation.
Mapping against Ray Data Co
The Topic → Subtopics → Proven Approaches hierarchy is directly assignable to Ray (the COO agent) as a batching task: supply a topic (e.g., "AI agents in enterprise data teams"), receive 7 subtopics, confirm, then generate 10 questions per subtopic. Outputs feed Sanity Check, RDCO LinkedIn, and phData thought-leadership directly — removing the "what do I write about" friction without requiring founder ideation time. The prompt's pause-for-confirmation pattern also maps to how the agent loop should gate multi-step content operations before expanding.
⚠️ Sponsorship
PS-section promotes AI Writing Skool (https://aiwritingskool.com/) — Dickie and Cole's own paid community (625+ members, weekly live trainings). Self-promotion; no third-party sponsor.
Related
- [[2026-06-04-ship30for30-30-linkedin-prompts-4-buckets]] — 30 LinkedIn prompts sorted into 4 idea buckets; complementary input-level taxonomy to this output-generation system
- [[2026-07-08-write-with-ai-consistent-voice-writing]] — AI-assisted writing with consistent voice; pairs with this framework at the drafting stage after ideas are generated