Every LinkedIn prompt we have, and the 15 jobs none of them does
28 prompts, mapped onto the 46 jobs people actually use LinkedIn for. 26 of those jobs have a prompt written for them. 5 are an output of a prompt filed under a different job. 15 have nothing, and are printed as empty rather than filled with something written this afternoon to close the row. Everything here is text you paste: no extension, no automation, nothing that connects to your account.
6 disciplines, 46 jobs, and 15 rows with nothing in them
A library organised by its own categories tells you what it has. Organised by the work, it also tells you what it is missing — which is the answer you need first, because it is the one that decides whether to keep reading or go elsewhere. The job list below is ours, not LinkedIn's: it is the work people describe when they ask for a prompt. Read the third column of each table before the second.
A prompt exists whose whole purpose is that job. 26 of the 46 rows. The link goes to the full text, the variables, a worked example and the failure modes.
5 rows. The job gets done, but as one output of a prompt filed under a different job — the post writer returns three openings, so there is no separate hook prompt. The cell says which prompt to open.
One row. Not a gap waiting to be filled: the library will not ask an assistant to invent the material it is supposed to be working from, and a topic generator is exactly that.
14 rows with nothing. Each one says what the nearest thing here is and where it falls short, so you can decide in a line rather than by reading four prompt pages.
Sales
“I need to find people worth talking to and get some of them to reply.” — 11 of 12 jobs have a prompt of their own.
| Job | What you are trying to do | Where it is |
|---|---|---|
| ICP research | Write down who you sell to in words a LinkedIn filter accepts. | ICP Builder |
| Prospecting | Turn that definition into a search that returns a workable number of people. | Sales Navigator Search Builder, Boolean Search Generator |
| Timing | Know which week to reach out, rather than reaching out the week you built the list. | Buying Signal Analyzer, Saved Search & Alert System |
| Qualification | Cut the list to the rows worth an invitation before you spend one. | Lead List Qualifier |
| Account research | Work out who decides, at a company big enough that the first contact sets the deal. | Account Research Brief |
| Prospect research | Find the one thing about this person that makes a message worth answering. | LinkedIn Prospect Research |
| Connection request | Say something specific in under 300 characters, or decide to send no note at all. | Connection Request Generator |
| First message | The message after they accept — or the paid one, when they will not see a request. | LinkedIn Outreach Personalizer, InMail Writer |
| Follow-ups | A cadence with a stop rule, sized to what your account can send in a week. | Follow-up Sequence Builder |
| Objection handling | Read what a two-line reply actually means, then answer it without arguing. | Reply & Objection Handler |
| Call preparation | Be useful in the two minutes before a call on a four-week-old thread. | Meeting Prep Brief |
| CRM notes | Write the record after the call, in the shape your pipeline reviews expect. | No prompt Nothing here writes the record after a call. The meeting prep brief is the nearest thing and it runs before. A note-taker wired into your call software is better placed than any pasted prompt, because it has the transcript and this does not. |
All sales prompts on one page, in the order the work happens.
Content
“I want to post something worth being known for, without inventing statistics.” — 5 of 9 jobs have a prompt of their own.
| Job | What you are trying to do | Where it is |
|---|---|---|
| Voice | Get a first draft that sounds like you instead of four rounds of “less corporate”. | LinkedIn Ghostwriter — Project Instruction and BRAND-VOICE.md The Project instruction sets the voice for every post you ask for afterwards; the context file is the same description as a .md you attach once. |
| Idea generation | Be handed ten things to post about when you have nothing in mind. | Deliberately absent No prompt here does this, and none is planned. Both content prompts take material in — a call, an analysis, an internal note — and the post writer returns a verdict that there is not enough to publish rather than filling the gap. An assistant asked for ten LinkedIn topics with no input returns ten it invented, with invented numbers inside them. |
| Post writing | Turn something you already know into a post that opens well. | LinkedIn Post Writer |
| Hooks | Three different first lines for the same post, to choose between. | Inside another prompt LinkedIn Post Writer The post writer returns three alternative openings with every draft. There is no separate hook prompt, because a hook written without the post under it is a line you then have to write a post to justify. |
| Rewriting | Fix a draft you already have rather than start again. | Inside another prompt LinkedIn Post Writer Paste your draft as the material. The post writer treats it the same way it treats a call transcript, which is what a rewrite is. There is no standalone rewriter. |
| Comments | Say something under someone else’s post that is not “Great post!”. | Comment Strategy |
| Newsletters | Write an issue for a LinkedIn newsletter rather than a post. | No prompt Nothing. A newsletter is subscribed to rather than served by the feed, and both content prompts are written for the feed. The repurposing chain is the closest, because a newsletter issue is usually the long source the posts get cut from rather than the other way round. |
| Thought leadership | Decide what you want to be known for before you post for six months. | Founder Positioning Brief |
| Repurposing | Get a month of posts out of a webinar, a report or a long call. | Repurposing Chain |
All content prompts on one page, in the order the work happens.
Recruiting
“I have an open role and a finite number of InMail credits this month.” — 4 of 8 jobs have a prompt of their own.
| Job | What you are trying to do | Where it is |
|---|---|---|
| Job description | Write the posting the search and the outreach both refer back to. | No prompt Nothing. The shortlist prompt takes your requirement list as an input and will not invent one — the same refusal as the content side. A job description an assistant writes from a job title describes the title, not the role you are filling. |
| Candidate persona | Decide what the person you are looking for actually looks like. | No prompt Nothing. The talent map tells you where people with a given profile work; it does not decide what the profile is. That decision is the hiring manager’s and the library treats it as an input. |
| Boolean search | A string that finds the same job under every name it is given. | Boolean Search Generator Filed under Sales as well. It is the same prompt: skills and exclusions matter more than titles on both sides. |
| Sourcing | Find out which companies the people you need are currently at. | Talent Map |
| Screening | Get from forty profiles to a shortlist you can defend line by line. | Candidate Research & Shortlist |
| Comparison | Put two finalists side by side against the same requirements. | Inside another prompt Candidate Research & Shortlist The shortlist is ranked, with evidence per requirement and the gaps named, which is a comparison across the whole set. There is no standalone head-to-head prompt. |
| InMail | A message a passive candidate answers, on a credit you cannot get back. | Recruiter Outreach |
| Interview questions | The questions to ask this candidate, given what the profile did not say. | No prompt The shortlist prompt returns one screening question per candidate, aimed at the specific gap it found in that profile. That is a filter, not an interview kit, and nothing here is an interview kit. |
Job search
“I am applying steadily and hearing nothing back.” — 4 of 9 jobs have a prompt of their own.
| Job | What you are trying to do | Where it is |
|---|---|---|
| Targets and routine | Replace an application count with a list of places and a weekly habit. | Job Search Assistant |
| Profile audit | Know which three changes to your profile are worth making first. | LinkedIn Profile Optimizer |
| Headline | 220 characters that say what you do, not what you are called. | Headline & About Writer |
| About section | The three lines that sit above the fold before anyone clicks “see more”. | Inside another prompt Headline & About Writer The headline prompt writes both. They are one job in practice: the About opening has to finish the sentence the headline started, and two prompts produce two positions. |
| CV alignment | Rewrite the CV against one specific posting without losing the thread. | No prompt JOB-SEARCH.md Nothing reads a posting against a CV. The job-search context file keeps every draft — message, application, profile rewrite — arguing the same case about you, which is the part that usually breaks first. The per-posting edit is still manual. |
| Recruiter outreach | Message the hiring manager, the recruiter and a possible referrer. | Message a Recruiter or Hiring Manager |
| Networking | Ask for a referral from someone who barely knows you. | Inside another prompt Message a Recruiter or Hiring Manager Its third message is the referral ask, written for exactly that case — someone who has met you once and is being asked to spend their own credibility. |
| Job analysis | Read one posting and decide whether it is worth an application. | No prompt Nothing. The job search assistant builds the target list; it does not sit with one posting and rule on it. |
| Interview prep | Walk in knowing the company, the team and what they will ask. | No prompt Nothing. The meeting prep brief is written for a sales call and its closing questions are discovery questions — used for an interview it produces a briefing about the company and nothing about you. |
Networking
“I want to open a conversation that is not a sale.” — 0 of 5 jobs have a prompt of their own.
| Job | What you are trying to do | Where it is |
|---|---|---|
| Introduction | Ask a mutual connection to introduce you, or make the introduction yourself. | No prompt Nothing written for it. The connection request prompt produces the mechanics — one specific reference, under the character limit — but the reason it argues is a sale, so that line has to be replaced by hand. |
| Event follow-up | Message forty people you met at a conference while they still remember you. | No prompt Nothing. The window is days and the only material is a conversation you had, which is the one input none of these prompts asks you for. |
| Investor outreach | Open a conversation with an investor you have no warm route to. | No prompt Nothing. The founder positioning brief produces the claim you would open with and the not-list that keeps you off the wrong ground; the message itself is not in this library. |
| Partnership outreach | Propose working together to a company that is not buying anything. | No prompt Nothing. The outreach personalizer is the closest shape, and it argues a purchase. A partnership message argues a mutual reason, which is a different hypothesis and a different proof. |
| Founder networking | Build a circle of people at the same stage as you. | No prompt Nothing. The comment strategy is the one that transfers unchanged: being visible under someone’s posts before you message them is the same move whether the objective is a deal or a conversation. |
Agency
“I run this for several clients and the context keeps bleeding.” — 2 of 3 jobs have a prompt of their own.
| Job | What you are trying to do | Where it is |
|---|---|---|
| Client intake | Turn a kickoff call into the files the rest of the work reads from. | Agency Client Intake |
| Running several clients | Keep one client’s proof out of another client’s post. | Multi-Client Workflow |
| Client reporting | Send the month’s numbers with something said about them. | No prompt Nothing. The multi-client prompt sets up the folders, the Projects and the weekly routine; what you send at the end of the month is outside it. |
Two of the 6 disciplines have a page of their own — sales and content — where the prompts are laid out in sequence with the reasoning that connects them. The other four are this table and the prompt pages it links to. Where a prompt writes Sales Navigator filters, it writes them for the tier you hold: a Core seat is $119.99 a month, read from LinkedIn on 12 September 2026.
15 of the 46 jobs have nothing, and one of them never will
Collected in one place so the answer takes a screen rather than six tables. Every row names the nearest thing here and where it stops, because “nothing” is only useful if it comes with the reason. An entire discipline is on this list: networking has no prompt of its own, and the two that transfer are named below.
| Job | Discipline | What to do instead |
|---|---|---|
| CRM notes | Sales | Nothing here writes the record after a call. The meeting prep brief is the nearest thing and it runs before. A note-taker wired into your call software is better placed than any pasted prompt, because it has the transcript and this does not. |
| Idea generation | Content Deliberately absent | No prompt here does this, and none is planned. Both content prompts take material in — a call, an analysis, an internal note — and the post writer returns a verdict that there is not enough to publish rather than filling the gap. An assistant asked for ten LinkedIn topics with no input returns ten it invented, with invented numbers inside them. |
| Newsletters | Content | Nothing. A newsletter is subscribed to rather than served by the feed, and both content prompts are written for the feed. The repurposing chain is the closest, because a newsletter issue is usually the long source the posts get cut from rather than the other way round. |
| Job description | Recruiting | Nothing. The shortlist prompt takes your requirement list as an input and will not invent one — the same refusal as the content side. A job description an assistant writes from a job title describes the title, not the role you are filling. |
| Candidate persona | Recruiting | Nothing. The talent map tells you where people with a given profile work; it does not decide what the profile is. That decision is the hiring manager’s and the library treats it as an input. |
| Interview questions | Recruiting | The shortlist prompt returns one screening question per candidate, aimed at the specific gap it found in that profile. That is a filter, not an interview kit, and nothing here is an interview kit. |
| CV alignment | Job search | Nothing reads a posting against a CV. The job-search context file keeps every draft — message, application, profile rewrite — arguing the same case about you, which is the part that usually breaks first. The per-posting edit is still manual. |
| Job analysis | Job search | Nothing. The job search assistant builds the target list; it does not sit with one posting and rule on it. |
| Interview prep | Job search | Nothing. The meeting prep brief is written for a sales call and its closing questions are discovery questions — used for an interview it produces a briefing about the company and nothing about you. |
| Introduction | Networking | Nothing written for it. The connection request prompt produces the mechanics — one specific reference, under the character limit — but the reason it argues is a sale, so that line has to be replaced by hand. |
| Event follow-up | Networking | Nothing. The window is days and the only material is a conversation you had, which is the one input none of these prompts asks you for. |
| Investor outreach | Networking | Nothing. The founder positioning brief produces the claim you would open with and the not-list that keeps you off the wrong ground; the message itself is not in this library. |
| Partnership outreach | Networking | Nothing. The outreach personalizer is the closest shape, and it argues a purchase. A partnership message argues a mutual reason, which is a different hypothesis and a different proof. |
| Founder networking | Networking | Nothing. The comment strategy is the one that transfers unchanged: being visible under someone’s posts before you message them is the same move whether the objective is a deal or a conversation. |
| Client reporting | Agency | Nothing. The multi-client prompt sets up the folders, the Projects and the weekly routine; what you send at the end of the month is outside it. |
This list is the reason the map is worth having. It will get shorter, but not by being padded: a row is filled when the job has come up often enough to be worth writing for, and the prompt that fills it will carry the same version, date and stated limits as the 28 that are already here.
All 54 items in one list, because a search should not have to pick a tab
Prompts, context files, agent packs and workflows together, filterable rather than split. Someone searching “boolean” does not know whether the answer is a prompt, a file or a pack, and a row of tabs makes them guess before they are allowed to look. The format has not gone anywhere — it is the first filter row and the label at the top of every card.
- Prompt
LinkedIn ICP Builder Prompt
An ICP written in the vocabulary LinkedIn search understands, derived from deals you actually closed rather than from who you wish would buy.
- Prompt
LinkedIn Prospect Research Prompt
A one-page research brief on a single prospect that ends in a usable opening line, and that separates what it read from what it inferred.
- Prompt
LinkedIn Lead Qualification Prompt
Your list, split into three tiers with a one-line reason per row, sized to what a LinkedIn account can actually contact in a week.
- Prompt
LinkedIn Buying Signal Prompt
A ranked signal list for your specific product, with the LinkedIn surface each one appears on and the window it stays actionable for.
- Prompt
Sales Navigator Search Builder Prompt
A filter-by-filter Sales Navigator search you can build in about three minutes, plus the Boolean strings for the fields that accept them.
- Prompt
LinkedIn Boolean Search Generator Prompt
A copy-paste Boolean string built from real title variants, with the exclusions that remove the usual pollution.
- Prompt
Sales Navigator Saved Search Prompt
A named set of saved searches inside your plan’s limit, each with a watch reason, a check cadence and the action its alert triggers.
- Prompt
LinkedIn Account Research Prompt
An account brief that names the committee, picks the entry point and states one testable hypothesis — built only from what is visible on LinkedIn.
- Prompt
LinkedIn Connection Request Prompt
Three connection requests under 300 characters, each with one specific reference, plus a recommendation on whether to send a note at all.
- Prompt
LinkedIn Cold Outreach Prompt
A first message and its follow-up, built from a stated hypothesis, with every claim traceable to something you actually gave it.
- Prompt
LinkedIn Follow-up Sequence Prompt
A three-to-four touch sequence with dates, channels, a stop rule, and a capacity check against what your plan can actually send.
- Prompt
LinkedIn Reply Handling Prompt
A read of what the reply actually means and three responses at different levels of directness, including the one that ends it politely.
- Prompt
LinkedIn InMail Prompt
A subject line and body written for a paid message, preceded by an honest verdict on whether to spend the credit at all.
- Prompt
LinkedIn Meeting Prep Prompt
A one-page brief you can read in the two minutes before a call, ending in the three questions that decide whether it goes anywhere.
- Prompt
LinkedIn Profile Optimizer Prompt
A ranked list of profile changes with the rewritten text for the top three, judged against one stated goal instead of against a generic checklist.
- Prompt
LinkedIn Headline Generator Prompt
Six headline options across three strategies, with character counts, plus the three lines of the About section that sit above the fold.
- Prompt
LinkedIn Founder Positioning Prompt
A single defensible claim, three supporting themes, an explicit not-list, and a one-question test for whether any given post belongs.
- Prompt
LinkedIn Post Prompt
A post built from material you supplied, with three alternative openings and an honest verdict when there is not enough there to publish.
- Project instruction
LinkedIn Ghostwriter Project Instruction
A standing instruction you paste once into a Claude Project or ChatGPT Project, after which every post you ask for arrives in your voice and with a critique already applied.
- Prompt chain
LinkedIn Repurposing Prompt Chain
A set of independent posts extracted from one source, each standing on its own claim, with an honest count rather than a forced ten.
- Prompt
LinkedIn Comment Prompt
Three comments at different levels of directness, plus a verdict on whether this post is worth commenting on at all.
- Prompt
LinkedIn Candidate Research Prompt
A ranked shortlist with evidence per requirement, the gaps named, and a screening question per candidate.
- Prompt
LinkedIn Recruiter Outreach Prompt
A subject line and message aimed at one candidate, with the details that decide whether a passive candidate replies included rather than withheld.
- Prompt
LinkedIn Talent Mapping Prompt
A named list of source companies, the titles the role is called at each, and the search strings that find them.
- Prompt
LinkedIn Job Search Prompt
A target list, a positioning statement, and a weekly routine — instead of an application count.
- Prompt
LinkedIn Message to a Recruiter Prompt
Three messages — to the hiring manager, to the recruiter, and to a potential referrer — each under the connection note limit where it needs to be.
- Prompt
Agency LinkedIn Client Onboarding Prompt
A filled ICP, voice and product brief per client, plus an explicit list of what the kickoff failed to establish.
- Prompt
Agency LinkedIn AI Workflow
A per-client folder and Project structure, a weekly routine, and the isolation rules that keep the work clean.
- Context file
ICP.md
Every prompt you run afterwards knows who you sell to, in terms LinkedIn can filter on.
- Context file
LINKEDIN-LIMITS.md
Your assistant stops proposing 300 connection requests a week, because it now knows the ceiling is about 100.
- Context file
BRAND-VOICE.md
Drafts arrive in your voice on the first attempt instead of after four rounds of "less corporate".
- Context file
COMPANY.md
The assistant stops describing your company the way its training data describes companies in your category.
- Context file
PRODUCT.md
Messages stop describing features and start describing what changes for the person receiving them.
- Context file
OUTREACH.md
Every message an assistant writes comes out at the right length, in the right shape, with the openers you hate already banned.
- Context file
OBJECTIONS.md
The assistant answers the objection you actually get, in the way that has worked, instead of improvising.
- Context file
CLAUDE.md
An agent that reads the right files in the right order and knows what it is not allowed to do.
- Context file
AGENTS.md
Three narrow agents with defined handoffs instead of one conversation that drifts.
- Context file
RESEARCH.md
Research that is consistent between people and between runs, with the evidence line drawn in the same place every time.
- Context file
SALES-NAVIGATOR.md
Search instructions that reference filters your tier actually has.
- Context file
PERSONAS.md
Messages that speak to what the reader is measured on rather than to a job title.
- Context file
CONTENT-ENGINE.md
A posting system with a material pipeline, rather than a calendar that dies in week three.
- Context file
RECRUITER.md
Sourcing and outreach that stays inside your InMail budget and does not infer things it should not.
- Context file
JOB-SEARCH.md
Every draft — message, application, profile rewrite — argues the same case about you.
- Agent pack
LinkedIn Sales Agent
A working agent folder — root instruction, seven context files and the prompts — that you fill in once and use for a year.
- Agent pack
LinkedIn Recruiter Agent
A sourcing assistant that stays inside a 30-credit month and does not score things a profile cannot evidence.
- Agent pack
LinkedIn Content Agent
Drafts in your voice, built from real material, with a self-critique attached before you read them.
- Agent pack
LinkedIn Job Search Agent
Every message, application and profile rewrite argues the same case about you.
- Agent pack
LinkedIn Agency System
A per-client structure where one client’s proof point cannot end up in another client’s post.
- Agent pack
LinkedIn AI Starter Pack
The whole toolkit as one folder, set up in twenty minutes, whatever your role is.
- Agent pack
LinkedIn MCP Setup Kit
A folder you can wire an assistant to LinkedIn from, with the exposure of each route stated before the first run rather than after it.
- Workflow
From ICP to booked meeting
Eight steps from "we should do outbound" to a call in the calendar, with the prompt for each and the artefact it produces.
- Workflow
From raw material to a posting habit
A posting system fed by real material, so the account survives a busy month.
- Workflow
From open role to a shortlist worth defending
A shortlist with evidence per requirement and an outreach plan that fits your InMail allowance.
- Workflow
From applying into the void to conversations
A weekly routine aimed at hiring managers rather than at an application counter.
What the five controls narrow, and what the line at the bottom of a card means
The controls above the list are the whole navigation. Nothing in it is ranked by what gets clicked; the prompts come first, in the order the work happens, then the context files, the packs and the workflows.
Matches the title, the one-line outcome, the summary under it and the roles an item is written for — and on prompts, the situations listed under “use it when” as well. Describing your position works as well as naming a tool: try “no reply”, “connection request”, “first message”.
Job narrows to a pasted prompt, a .md context file, a zipped agent pack or a workflow that runs several in order. Task narrows to an area of work. Reach for Task when you know what you are doing but not which thing does it.
Who the item was written for. Items carry every role they genuinely serve rather than being filed under one, so a prompt written for both an SDR and an agency appears under both.
One item is marked “Free, email required”: the Starter Pack zip, where the form asks for an address so we can tell you when LinkedIn moves the numbers those files are built on — and the ungated download sits directly under the form anyway. Everything else is free with no account, and the Paid chip returns nothing, because nothing here is paid.
Not a difficulty rating and not a runtime. It names the manual work the item is meant to displace — “Replaces the 10–15 minutes of tab-hopping before a first message” — because across a list this long that is the comparison worth making. On a workflow it is the step count instead.
A filtered view is not a page: it has no distinct intent and would produce hundreds of thin crawlable permutations of this list. The canonical answer to “prompts for recruiters” is the recruiting section of the map above, which has copy on it.
Open these five, in this order
Both a map and a filter assume you know what you are looking for. If you do not, start with the opening five steps of From ICP to booked meeting. Each one consumes what the one before it produced, which is why the order matters: run them out of sequence and they are five unrelated prompts.
- LinkedIn ICP Builder Prompt — define who. What it hands the next step: An ICP in LinkedIn’s own vocabulary, plus one observable trigger.
- Sales Navigator Search Builder Prompt — find them. What it hands the next step: A filter set and Boolean strings returning 200–1,500 results.
- LinkedIn Lead Qualification Prompt — cut the list. What it hands the next step: Three tiers, capped at what your weekly invitation budget can actually cover.
- LinkedIn Prospect Research Prompt — understand the top tier. What it hands the next step: A brief ending in a value hypothesis and an opening line.
- LinkedIn Cold Outreach Prompt — write the first touch. What it hands the next step: A message at the right length, built on a stated hypothesis.
The other 3 run the same way for content, recruiting and job search: all 4 workflows.
Five the individual prompt pages do not answer
Where an answer rests on a LinkedIn number, the number is not repeated here. It lives on the LinkedIn limits page, dated, so there is one place to correct when LinkedIn moves it.
Will using any of these get my LinkedIn account restricted?
Nothing here touches your LinkedIn account. Every item is text — a prompt you paste, a .md file you attach to a Project, or a workflow that runs several in order — and what comes back is a message, a filter set or a brief that you then use yourself in LinkedIn’s own interface. There is no extension, no automation and no API call, and no prompt here asks an assistant to open a LinkedIn page — it cannot, because LinkedIn blocks assistant crawlers, and a model handed a profile URL invents the contents instead. What does get accounts restricted is volume: sending more invitations or messages in a week than the platform allows. That is why the prompts carry the ceilings rather than leaving the arithmetic to you: a list of four hundred people to contact this week is not an ambitious plan, it is a plan that gets the account restricted.
Why print the jobs nothing here does, instead of writing prompts for them?
Because a library that covers everything is a library you cannot trust about anything. Writing 15 prompts this afternoon to close the empty rows would produce 15 prompts nobody had a reason to write, and it would destroy the only thing the map is for — telling you in one screen whether to keep looking. Some of the gaps are also deliberate rather than pending: no prompt here generates post ideas from nothing, and no prompt writes a job description from a job title, because an assistant given no material invents it. When one of these rows is filled it will be because the job came up often enough to write for, and the row will say so.
Should I use a prompt or a context file for the same job?
Ask how often you will do the job. A prompt is consumed by one message and dies with the conversation; a context file is read on every turn for as long as it stays attached. One profile to rewrite, one reply to answer: paste the prompt into a blank chat and be done. Anything weekly: put the files in a Project once, and every prompt you run there inherits your ICP, your product and the limits without you retyping them — which also makes the prompts shorter and stops two answers contradicting each other about who you sell to. They are not really alternatives. The prompts that expect files name them on their own pages.
Do these work with Claude, ChatGPT and Gemini?
Yes, and most of them are not written for one of the three. A prompt with a role, stated constraints and an output schema is not model-specific, and splitting one into a Claude version and a ChatGPT version that differ in the first line is how prompt sites end up with two thin pages competing for the same query. An item is filed under one assistant only where the setup genuinely differs — a Claude Project with attached context files is a different object from a pasted chat prompt. What the library claims about testing is narrow on purpose: each item was written to a fixed structure and run through Claude and ChatGPT to confirm it returns the output format it declares. That is not a measured reply-rate benchmark and nothing here says it is.
Why is there a map and a flat list rather than one or the other?
Because two different people arrive here. One has a job — “message a passive candidate”, “fix my headline” — and wants to know whether this library has it, which a search box cannot answer when the answer is no. The other has a keyword, usually one that does not respect the format boundary: someone searching “boolean” does not know whether the answer is a prompt, a context file or a pack, and a row of tabs would make them guess before they are allowed to look. The map answers the first and prints its own holes; the flat list answers the second across all 54 items at once. Neither is a filtered view of the other, and neither gets its own URL — a filtered list is not a page.
Find the job first, the prompt second
26 of the 46 jobs on the map have a prompt written for them and 15 do not. Checking which yours is takes one screen, and it is a faster answer than any search box can give you.