Running prompt library

Prompts worth using twice.

This page collects the prompts that create a useful artifact, not just a nicer chat reply. Use them in Claude Code, adapt them to your work, and send the ones that save you real time.

Prompt shelf
Output Open the right drawer, copy the prompt, run it on real work.
Coach pickAnti-sycophancy
LibraryCoding, extraction, strategy

Prompt shelf

Pick the job first. Then open the prompt.

This shelf is organized by what you are trying to do: sharpen the model's judgment, build cleaner code, extract useful material, or run a strategy pass. The full prompt only appears when you open a drawer.

1 prompt

Model behavior

Use these when you need Claude to push back, inspect the work, and stop rewarding your framing.

Anti-Sycophancy PromptFor strategy reviews, big decisions, and copy critique. Guardrail

Use this when you need disagreement to be real, not polite validation.

Open this drawer to load the prompt.
3 prompts

Coding checks

Use these before shipping a site, app, or feature that needs to survive a real review.

Mobile Responsive Polish PassRun after the desktop version works, before deployment. Mobile

Use this to catch stacking, tap target, type scale, and overflow problems before a page goes live.

Open this drawer to load the prompt.
Read-Only Disk Bloat AuditFind what is filling your machine before deleting anything. Maintenance

Use this when Claude Code, Codex, build folders, caches, logs, or generated media have started eating local disk space.

Open this drawer to load the prompt.
System Architecture RulesPaste before a coding session when the repo has no local rules. Architecture

Use this to force the model to inspect existing routes, tables, data stores, and ownership before it starts building.

Open this drawer to load the prompt.
1 prompt

Surface builds

Use this when the page has to sell, not just exist. It locks the buyer, promise, mechanism, proof, and above-the-fold argument before Claude Code builds.

Landing Page Build CoachRun before the Surface Brief when the page is a sales page. Day 3

Use this after Surface Scout and before `SURFACE-BRIEF.md`. It keeps the page from turning into purple fog with three generic cards.

Open this drawer to load the prompt.
3 prompts

Extraction

Use these when raw material has value, but the useful structure is still buried in the source.

Modular Content ExtractorFind frameworks, repeatable moves, and productizable blocks. IP blocks

Use this on transcripts, brainstorms, methodology notes, and client work artifacts.

Open this drawer to load the prompt.
Post-1-on-1 Employee ExtractionTurn a meeting transcript into structured case-study data. Transcript

Use this when a 1-on-1 contains before state, friction, proposed build, time saved, and adoption signals.

Open this drawer to load the prompt.
Content-to-Prompt ScannerIdentify rare material worth turning into reusable prompts. Scanner

Use this on articles, podcasts, videos, or transcripts when you want only high-confidence prompt opportunities.

Open this drawer to load the prompt.
4 prompts

Strategy frameworks

Use these when you need a thinking pass with a named mechanism, not a loose brainstorm.

Compound Learning: 50/50 Meta LayerExtract what the project taught you before the lesson evaporates. Learning

Use this after a build or deliverable so the next similar project starts ahead.

Open this drawer to load the prompt.
Customer Journey Gap AnalyzerMap where customers are lost, under-served, or under-converted. Journey

Use this when the funnel looks active but the path from attention to retention has weak points.

Open this drawer to load the prompt.
Double Down DiagnosticFind what is already working and what should stop getting attention. 80/20

Use this before adding a new strategy when the current business already has a hidden advantage.

Open this drawer to load the prompt.
Viral Amplifier MatcherMatch psychological triggers to an existing strategy or content system. Amplifier

Use this when the base system works, but the attention mechanism needs sharper hooks.

Open this drawer to load the prompt.

Community shelf

Send the prompt that made your work easier.

Accepted prompts are posted with the person tagged. Include the handle exactly how you want it shown.

What gets posted

Useful, repeatable prompts

A prompt belongs here when another person in the course could paste it, swap the specifics, and get a better artifact.

What gets rejected

Private data or vague tricks

No client secrets, private transcripts, account screenshots, or prompts that only work because of hidden context.

Attribution

Name, handle, and source

We keep the submitter attached to the prompt. If it came from a public post, include that source too.


Send

Send a prompt to the library.

Version 1 captures submissions for coach review. That keeps spam, private client material, and broken prompts off the public page. A true auto-post feed needs a small backend with moderation rules before it should go live.


Coach review

The posting workflow.

This is the admin surface for now: clear enough for coaches to run without needing a separate tool.

  1. Read the submission

    Check the prompt, use case, source link, and requested attribution.

    If the context is missing, ask before posting.
  2. Remove private material

    Reject client names, private links, screenshots, transcript excerpts, or hidden business data.

    The library should teach the move, not expose someone else.
  3. Test the prompt once

    Run it on a harmless example and make sure the result is useful.

    A prompt that sounds good but fails in practice should not ship.
  4. Post with attribution

    Add the prompt, the submitter tag, source link, and one sentence on when to use it.

    The tag gives credit and makes good prompts social.