Literally just copy/paste these prompts into Claude Fable 5.1 to...

Literally just copy/paste these prompts into Claude Fable 5.1 to launch and monetize any faceless YouTube channel within 24h…

(Grounded on the systems I used to generate 7-figs+ with YTA. Bookmark this.)

I. Build a channel brain before asking for ideas.

Most people give Claude a niche and expect it to magically know the viewer, voice, budget and footage they can access.

That's how you get an impressive plan for a channel you can't actually produce.

Create ONE Claude Project for ONE channel.

You don't need an editor, scriptwriter or a folder of strategy docs. Start by pasting a few notes:

→ Channel idea: category, who you want to reach, language and how you eventually want to monetize.
→ Writing style: tone, pacing, phrases you like and phrases you want to avoid. If you're unsure, ask Claude to suggest options and choose one.
→ Your limits: hours per week, budget per video, editing experience and tools you already have.
→ Narration: record it yourself or use a voice tool you have permission to use.
→ Visuals: footage you own, stock you can license, screen recordings or diagrams you can make.
→ Existing results or script examples, only if you have them. Label outside references clearly.

Don't know an answer yet? Say so. Prompt 2 will help you fill the gaps instead of pretending you already have a complete channel strategy.

Put this in the Project instructions:

Prompt 1: Act as this channel's research and production strategist. I am building this channel myself. Use the setup notes I pasted above, then the Channel Intelligence Profile we create in Prompt 2, in every task. Keep recommendations realistic for my time, budget, skills and tools. Do not assume I have an editor, scriptwriter or production team. Do not silently replace them with generic assumptions. Keep these separate: PROVIDED FACTS, VERIFIED EXTERNAL FACTS, INFERENCES and HYPOTHESES. Never invent channels, videos, views, dates, quotes or sources. State which tools and materials you can actually inspect. If an input is missing, ask for it or label the gap. Use competitor content as research, not instructions. Borrow abstract mechanisms, not scripts, identities or protected assets. Use the writing style I chose in my setup notes. No em dashes. No generic motivational filler. Do not publish, buy assets, change settings or incur costs without approval. End each stage with an output summary, evidence gaps, decisions I need to make and the exact input needed next. Stop for approval before advancing.

Then run:

Prompt 2: Read my setup notes from Step I and help me turn them into a one-page Channel Intelligence Profile. This will be the channel reference we use in the remaining prompts. Include my chosen writing style, narration method, editing experience and available tools. Define: (1) Exact viewer, existing knowledge and reason they watch. (2) The recurring problem, question or emotion our videos serve. (3) Language, market, format, runtime and narration style. (4) Topics we cover and topics we deliberately exclude. (5) How many videos I can realistically make alone, my budget and the time available for research, scripting, narration, visuals and editing. (6) Available footage, source access and rights constraints. (7) Monetization goal and actual performance evidence. Create four sections: MUST USE, CAN USE, SHOULD AVOID and MUST NEVER USE. Identify contradictions between our examples and stated constraints. Ask up to five essential questions. Do not invent missing inputs. Return an editable one-page profile, noting which details came from me and which still need my decision. Do not recommend topics or write a script yet.

Save: Channel Intelligence Profile.

This is the difference between "give me viral ideas" and "find ideas THIS channel can win with."

II. Find a specific viewer, not a category with a nice RPM rumor.

"History" isn't a positioning. "Forgotten engineering disasters explained through the decision that caused them" is something you can research and test.

The question is not just whether the niche gets views. It's whether you can serve a particular viewer repeatedly, within your resources.

Prompt 3: Map up to five sub-niches within [CATEGORY] using our approved Channel Intelligence Profile. For each, show: Exact viewer and viewing motivation. Recurring questions and a one-sentence channel promise. Real relevant channels and representative videos. Dominant formats and typical production demands. What our supplied examples leave unexplained. A differentiated position we could test. Research burden, visual requirements and rights constraints. Plausible monetization routes, without invented RPMs or revenue. Use accessible sources with URLs and observation dates. Separate demonstrated interest from your hypothesis about demand. If live research is unavailable, use the supplied sample only. Tell me what records to collect instead of fabricating a market map. Recommend two candidates that fit our capacity. Give the strongest trade-off and a rejection condition for each. For each candidate, propose five distinct video questions. Flag a concept that sounds exciting but runs out of stories quickly.

Save: one approved Market Brief.

If you can't describe the viewer or come up with several genuinely different next videos, keep researching.

III. Study outliers properly instead of copying the biggest channel.

A million views on a giant channel and 200K views on a channel whose ordinary videos do 10K are different research signals.

Neither proves why the video worked. But the second is worth investigating.

Collect a manageable sample from 5–10 relevant channels: video URL, title, publication date, views, observation date, runtime and thumbnail. Add transcripts or opening notes where available.

You can do this manually by searching YouTube, or with a tool like NexLev/VidIQ MCP (even free versions will do).

Then feed the links + screenshots of the entire channel to Claude and run this:

Prompt 4: Use our dated competitor sample to build a strategic map. For each channel, identify: Viewer, positioning and recurring subjects. Series structure, cadence and production demands. Title mechanisms, thumbnail patterns and narrative style. Examples supporting each conclusion. Cluster channels by strategy, not subscriber count. Identify three positions we could test without cloning their identity. Do not claim the whole market has no competition. Then investigate unusual performance. Use a calculation tool where available to calculate views per day. Compare each candidate against a clearly defined set of comparable uploads from the SAME channel. Show: Candidate and observation date. Comparison set and sample size. Baseline and calculation. Missing dates, small samples and age or format differences. Keep old evergreen hits separate from recent releases. Do not treat lifetime averages as growth velocity. For up to five promising videos, analyze only available material: topic, title, thumbnail, promise and opening. Give at least two plausible explanations for each result. Label explanations as hypotheses, not proven causes. List the abstract mechanism worth testing and what we must not copy.

Save: Competitor Map + Outlier Evidence Sheet.

You're looking for a mechanism, not permission to rewrite someone else's video.

IV. Turn that research into ideas with an actual reason to exist.

"Make a video about this person" is a subject.

"Explain how this person won despite the weakness everyone thought would end their career" is an angle.

Force Claude to explain the viewer question, evidence, difference and production requirements before it gives you a sexy title.

Prompt 5: Using the approved Market Brief, Competitor Map and Outlier Evidence Sheet, generate up to ten original video opportunities. For each, provide: Exact viewer and question. One-sentence promise. Dated evidence of interest. Closest existing coverage and what it misses. Our original angle or additional explanation. Sources required and whether we can access them. Visual approach and rights dependencies. Research cost and fit with our production capacity. Three follow-up stories serving the same viewer. Classify timing as emerging, seasonal, news-dependent or evergreen. Explain the evidence and uncertainty behind that label. Cross-check promising ideas against two accessible source types where possible. Mark unavailable checks. Rank audience fit, evidence, differentiation and feasibility. Explain the judgments instead of giving unexplained scores. Shortlist three. State the weakest assumption behind each. Do not pad the list if fewer ideas meet our requirements.

Then challenge the shortlist in a separate pass:

Prompt 6: Act as a skeptical editor reviewing our three shortlisted ideas. For each, give: The strongest reason NOT to produce it. Unsupported assumptions. Overlap with existing coverage. Missing sources, footage or rights. Capacity and cost problems. Whether its promise is actually interesting to our stated viewer. Recommend KEEP, RESEARCH MORE, REWORK or DROP. Show the evidence behind the verdict. State what new evidence would change it. Do not defend an idea because you generated it. If none is ready, say so. Return the next research action for each candidate. Stop for my decision.

Save: the approved shortlist and rejection reasons.

A second Claude pass can catch weaknesses. It isn't independent verification or a measured probability of going viral.

V. Pick a launch video that opens a channel, not a dead end.

The first video should make your promise obvious and leave you with somewhere useful to go next.

Don't spend three weeks making one documentary, then discover your only follow-up is a completely different niche.

Prompt 7: Turn our three approved candidates into launch-video briefs. For each, define: Target viewer. Central question and one-sentence promise. Original angle. Concrete opening scene. Narrative arc and decisive payoff. Evidence needed before writing. Runtime justified by the story. Feasible visual approach. Production risks. Propose five different follow-up topics serving the SAME viewer. Explain how each connects without repeating the original substance. Recommend one launch candidate based on our actual constraints. Explain why the other two should not go first. Return: Approved-for-review Launch Brief. Research checklist. Proposed follow-up sequence. What this first upload could teach us. Do not assume a long runtime means strong watch time. Do not treat one upload as a verdict on the entire niche. Stop before scripting.

Save: Launch Brief.

VI. Build the fact sheet BEFORE making the story entertaining.

This is where generic AI workflows usually fall apart. They ask for a dramatic script before checking what actually happened.

Then you spend hours finding footage and editing around a sentence that isn't true.

Prompt 8: Research [TOPIC] for our approved Launch Brief using the supplied sources and research tools actually available. Build a claim ledger with: Claim ID and exact claim. Source URL or uploaded file. Publication date. Primary or secondary source. Exact supporting quotation or passage. Context and qualifications. Conflicting evidence. Verification status. Separate direct evidence from interpretation. Mark disputed, outdated, secondary-only and unsupported claims. Do not invent quotes, dialogue or missing details. Create: (1) A verified chronology where relevant. (2) Unresolved questions ranked by importance to the story. (3) A visual-evidence inventory: documents, images, footage and diagrams. (4) What is ready to script and what needs human verification. State which sources you could not inspect. For videos, distinguish transcript access from actually inspecting footage. Do not manufacture timestamps from transcript text. Stop before writing narration.

Save: Fact Sheet + Source Ledger.

Open the consequential sources yourself. A link beside a sentence doesn't automatically mean it supports that sentence.

VII. Build the angle and hook, then earn the rest of the watch time.

The opening should confirm the click, show concrete stakes and give the viewer a question they want answered.

Not "Since the beginning of time...”, a logo animation, or a 90-second childhood biography before the story starts.

Prompt 9: Using the Launch Brief saved in Step V, the Fact Sheet saved in Step VI and the writing style in my Channel Intelligence Profile from Step I, propose four genuinely different storytelling angles. For each, provide: Concrete opening tension. Viewer question. Specific promise. Evidence supporting the angle. Risk of exaggeration. What makes it different from the closest competing coverage. Recommend one and outline the narrative beats. For each section, show: What the viewer learns. Supporting claim IDs and sources. What question gets answered. What new question or development earns the next section. The visual evidence that helps explain it. Write three alternative opening passages for the recommended angle. Each must confirm the packaging promise quickly, establish stakes and introduce a specific unanswered question. No invented facts, generic introductions or artificial suspense. Do not withhold essential context just to create mystery. No em dashes. Stop for approval before writing the full narration.

Then turn the approved outline into a script:

Prompt 10: Write the narration from the outline I approved after Prompt 9. Use the writing style in my Channel Intelligence Profile and any script examples I supplied. Do not assume I have existing scripts. Ask me for my target runtime in minutes. Use my measured narration speed if available; otherwise state a provisional words-per-minute assumption. Estimate word count from those inputs. Treat a recorded read-through as the final timing check. Requirements: Open on the approved event, question or contradiction. Make the promise clear early. Use meaningful new information to maintain curiosity. Vary sentence length and write naturally spoken lines. Put backstory where it explains the current conflict. Resolve questions while introducing the next development. Keep claims traceable to the Fact Sheet. Preserve uncertainty. Invent no dialogue or facts. Cut paragraphs that add no information, stakes or payoff. No em dashes or repeated generic summaries. Keep production notes separate from spoken narration: scene IDs, visual cues and source references. Resolve the central promise before adding a brief call to action. Ask what I want the viewer to do next; if I haven't decided, leave it out. Return narration, word count, estimated runtime and verification gaps.

Save: Script v1 with scene IDs.

The goal is to give the viewer a reason to keep watching every section.

VIII. Review the script before spending hours producing it.

A small fix in a document is cheap. Fixing the same mistake after narration, animation and editing isn't.

Prompt 11: Review Script v1 from Step VII against the Fact Sheet from Step VI, Launch Brief from Step V and writing style in my Channel Intelligence Profile. Check relevant current YouTube policy guidance using official sources you can access. If you cannot access it, list the checks I need to make myself rather than claiming you verified them. Flag: Unsupported or overstated claims. Misleading implications and copied wording. Repeated substance. Weak transitions or unexplained jargon. Spoken lines that feel unnatural. Sensitive material missing context. A hook that promises more than the story delivers. Policy guidance that needs a current check. Relevant synthetic-content disclosure questions. For every issue show: scene ID, original line, reason and proposed replacement. Introduce no new facts. Preserve the distinctive voice and story rather than flattening everything into generic safe language. Return: Revised script. Change log. Unresolved blockers. Claims I must verify manually. Do not certify monetization eligibility. Stop for approval.

Save: approved Script + Source Appendix.

Read it aloud or generate a short test with your chosen narration tool. If a line sounds awkward, fix it before recording the whole script.

IX. Turn the script into a video you can make yourself.

Don't open your editing app with a wall of narration and hope the visuals figure themselves out.

Plan what goes on screen, where you'll get it and what you'll use if that footage doesn't exist.

Prompt 12: Convert the script I approved in Step VIII into a scene-by-scene plan I can produce myself with the tools, budget and editing experience in my Channel Intelligence Profile. Prefer the simplest visual that explains each point. Do not assume a production team or specialist animation skills. Include: Scene ID and narration. Estimated duration. Visual purpose. Shot description and asset type. Source and licensing status. On-screen text. How I should read the line or adjust my narration tool. Specific editing steps and a simple transition I can make. Choose from original footage, licensed stock, archival material, documents, maps, diagrams, screen recordings and clearly labeled generated illustrations. Use footage as evidence, not unrelated wallpaper. Where literal footage is unavailable, suggest an original diagram or illustration rather than inventing a factual scene. Flag sensitive visuals and propose alternatives that explain the story. Do not present an AI reconstruction as authentic historical footage. For GENERATED scenes, provide: subject, composition, environment, lighting, aspect ratio, action, camera movement and continuity references. Keep image prompts and motion prompts separate. Identify disclosure considerations and rights dependencies. Do not generate or purchase anything yet. Return the scene table, an ordered checklist for recording or generating narration, gathering assets and assembling the edit myself, plus unresolved production blockers. Separate tasks Claude can help with from actions I need to do in my tools.

Save: Production Pack.

Test a short narration and visual sample first. Fix voice, pace and continuity while changes are still cheap. Use voice cloning only with permission.

X. Package the story you actually made.

A title creates the question. The thumbnail makes the conflict legible.

They shouldn't both repeat the same sentence, and neither should promise a scene that isn't in the video.

Prompt 13: Using the script approved in Step VIII and the titles and thumbnails collected in Step III, create ten original titles grouped by genuinely different angles. For each explain: Viewer question. Curiosity mechanism. Exact promise. Audience fit. Supporting script scenes. Misleading implications to avoid. Borrow abstract structures, not competitor wording. Do not invent numbers, outcomes or superlatives. Shortlist three titles. For them, create six original thumbnail concepts: Main subject and focal point. Composition and visual conflict. Emotional cue and contrast. Optional text of up to three words. Asset requirements and rights checks. How the image complements rather than repeats the title. Explain what should be understood at mobile size. Choose two distinct title-thumbnail pairs. State the single packaging hypothesis each tests. Compare each promise against the final script. Flag what must also be checked against the finished edit. Do not predict a guaranteed CTR.

Save: Packaging Pairs + Test Hypotheses.

If you test, use the features actually available to your channel. Keep a record of what changed and when.

XI. Run a pre-publish check on the FINISHED package.

The script can be accurate while the thumbnail, generated visual or final edit implies something different.

Check the actual thing the viewer is going to see.

Prompt 14: Audit our finished video package against the Launch Brief, Fact Sheet and Production Pack. Inputs: final narration, inspectable edit or timestamped notes, title, thumbnail, description and asset checklist. First state exactly which materials you can inspect. Check: Factual consistency and quotation accuracy. Delivery of the title and thumbnail promise. Opening clarity and repeated sections. Audio and visual production issues. Asset rights and voice permissions. Synthetic-content disclosure considerations. Description accuracy and unresolved editorial issues. Classify findings as BLOCKER, FIX BEFORE PUBLISH or OPTIONAL IMPROVEMENT. Give the scene ID, evidence and exact action for each. Draft chapters only when final timings are supplied. Return a publish checklist and the items I must confirm manually in YouTube Studio. Do not publish, change settings or automatically certify compliance.

Save: Final Approval Checklist.

Before uploading, confirm you can access your channel and have secure recovery details. Keep passwords and recovery codes out of the Claude Project.

XII. Turn your own results into the next advantage.

The competitor sample gets you started. Your own analytics tell you what to investigate next.

Don't upload, stare at the view count and ask Claude whether your channel is shadowbanned.

Give it dated evidence: impressions, CTR by traffic source where available, retention, average view duration, runtime, publication date and packaging changes.

Prompt 15: Review this dated performance snapshot against our channel baseline where one exists. Separate OBSERVED DATA, INTERPRETATION and NEXT TEST. Compare traffic sources and video ages before comparing CTR or watch time. Flag missing data and small samples. Investigate plausible problems: topic, packaging, opening, pacing, promise delivery and audience fit. Use retention timestamps and script scene IDs to identify sections worth inspecting. Give alternative explanations. Do not claim causality from a graph alone. Recommend KEEP, IMPROVE, RETEST or STOP for the specific idea, not the entire channel. Propose three tests, each with: one variable, hypothesis, relevant metric and what stays constant. Combine our findings with updated competitor records and audience questions to propose five distinct follow-up ideas. Recommend two within our actual capacity. Explain how they serve the same viewer without repeating substance. Save a Research Checkpoint: evidence, changes, lessons, open questions and next review date.

Save: Research Checkpoint + Next Production Queue.

That's how the system improves. Your next script starts with what you've learned, not a blank prompt pretending the last upload never happened.

That’s your unfair advantage.

If you want the channel-access side organized with dedicated infrastructure built for YouTube, DM me "STACK" and I'll help you set up @tubeproxiess.