Your Library

Everything behind the Dashboard link. How to manage your prompts and building blocks, collaborate with the community, and track how your work evolves.

1. The dashboard

The dashboard is your home for everything you have created and saved. You reach it at /dashboard when you are signed in. A sidebar on the left gives you direct access to each entity you can manage:

• Prompt builder - the same builder as the home page, available as a permanent dashboard entry.
• Prompts - every saved prompt you own, plus the community's.
• Libraries - curated groups of prompts you can keep private or share publicly. See section 14 for details.
• Queues - ordered lists of prompts designed to be worked through programmatically by agents. See section 15 for details.
• Users - the public reputation leaderboard, ranking community members by the votes and favorites their work has earned. See section 19 for details.
• Notifications - activity on your content: comments and remixes. A badge on the sidebar entry shows your unread count.
• Personas, Tones, Output Formats, Constraints - one page for each building block.
• API Keys - manage keys for programmatic access.
• Settings - your account, plan, and usage. Displays your current Reputation score alongside your content counts. Includes a Clear AI suggestion learning action that resets the personalization data used by the building-block suggester. See section 19 and section 20.
• Documentation - external link to this documentation site.
• API Documentation - external link to the REST API reference.
• NPM Package - external link to the official @prompty-tools/core package.

2. Mine, Public, Favorites, and Collections

Every entity page in the dashboard is organized into tabs:

• Mine - items you created. You can edit, rename, change visibility, and delete these. Deleting is permanent.
• Public - items the community published. You can view, vote, and favorite these. You cannot edit them.
• Favorites - everything you favorited across the community and your own work.
• Collections - (Tones and Constraints pages only) collections that group tones or constraints together. Tones and Constraints maintain separate collection libraries; see section 13 for details.

Prompts, Personas, Output Formats, and Libraries pages have three tabs (Mine, Public, Favorites). Tones and Constraints pages have a fourth: Collections. The Queues page does not use this tab structure — it shows only your own queues, which are always private.

The Mine tab shows usage counters while you are on a free plan, so you always know how close you are to the per-tier limit. See the Tiers & Access page for the exact numbers.

3. Creating and editing

Creation forms live inside each entity page, and inside the prompt builder itself. You never have to switch pages to create a block mid-build. The forms validate your input immediately.

• Character limits are enforced as you type, with a live counter.
• Tags are checked for shape (lowercase alphanumeric with hyphens) and length.
• Duplicates are caught at save time.

Editing a non-versioned entity (Tone, Output, Constraint) updates it in place. Editing a versioned entity (Prompt, Persona) creates a new version. Your prior version stays available in the version history.

4. Public and private

Every entity has a visibility: public or private. Visibility is set when you create the entity and can be toggled at any time from its detail page. The two modes differ in a few ways.

• Public items appear in the community's Public tab. They are voteable, favoriteable, shareable by URL, and visible to signed-out visitors. They do not count against any limit.
• Private items are visible only to you. They count toward the private-item limit on your plan.

Every entity defaults to public when you create it through the builder or the dashboard form. If you want a private item, tick the Private box before saving. You can flip visibility at any time from the item's detail page.

5. Tags and search

Tags help you organize and rediscover your work. Every entity supports up to 10 tags. Each tag is lowercase alphanumeric with hyphens, between 3 and 24 characters. Duplicates are removed automatically.

Search is available on every list. It matches the item's title, description, text, and tags, and it works across Mine, Public, and Favorites independently. Typing a tag value into the search box surfaces every item carrying that tag, even if the word does not appear in the title or body. Search and tag-filter pills can be combined: you can narrow by a tag pill and still type a free-text search inside that filtered view.

6. Sorting and pagination

Every list supports three sort orders: Newest, Most Upvoted, and Most Favorited. Dashboard lists paginate with configurable page sizes. In the builder, each building-block section loads more items inline with a Show More button.

7. Voting

Every public entity can be voted on. Voting is simple: thumbs up, thumbs down, or no vote. You can change your mind. Your vote is private; only the aggregate score is shown.

• Net score is upvotes minus downvotes. Both counts are visible on detail pages.
• Voting works across all six entity types. A persona, a tone, an output, a constraint, a prompt, and a collection can all be voted on.
• You can also vote on comments, which drives the Highest Rated sort order in the comment list.

Sorting a list by Most upvoted surfaces the work the community has validated. It is a good place to find patterns worth reusing.

8. Favorites

Favorites are your private shortlist. Click the heart icon on any entity to add it to your Favorites tab for that entity type. Click again to remove it.

• Favorites are per entity type: you have separate lists for favorite prompts, personas, tones, outputs, constraints, and collections.
• Favorites are subject to a per-tier limit. On the free plan you can have up to 5 favorites per entity type.
• The builder reads from your favorites automatically. Pills in the Tone and Constraints sections surface favorited items first.

9. Versioning

Prompts and Personas are versioned. Every edit creates a new, immutable version with its own compiled text, tags, and optional changelog. Older versions remain accessible forever.

• Changelog: up to 300 characters per version. A short note on what changed and why.
• Version history: every prompt and persona detail page has a link to its version history.
• Create from any version: you can open an older version and use it as the starting point for a new version.
• Tier limit: the number of versions per prompt is capped on the free plan. See Tiers & Access for the current limit.

Use the changelog

A one-line changelog explaining why you made the change will save you an hour next month when you are trying to remember which version to reuse. Treat it as a note to your future self.

10. Remixing prompts

Remix lets you use any prompt you can see as the starting point for a new one of your own. Instead of editing the original, which you cannot do unless you own it, you get a fresh copy in the builder pre-filled with everything from the source's latest version: the task, persona, tones, output, constraints, title, description, and tags.

• Where to find it: every prompt detail page in the dashboard has a Remix button next to Copy. You can remix your own prompts as well as any public prompt in the community.
• Building blocks: references to the source's persona, tones, output, and constraints are kept if those blocks are yours or public. If a referenced block was private to the original author, its text is still copied into your new prompt's snapshot, but the reference is dropped so later edits to that hidden block will not flow through.
• Saving: a remix is saved as a new prompt, not a new version of the source. You pick your own title, description, tags, and visibility. The new prompt only counts toward your private-prompt limit if you choose to make it private.
• Attribution: your new prompt's detail page shows a Remixed from link pointing at the exact source version, so anyone reading it can trace the lineage.
• Notification to the source author: when you remix someone else's public prompt, its author receives a notification so they can see what you built with their work.

Use remix instead of starting over

If a community prompt is almost what you want, remix it, adjust the task and a constraint or two, and save. You keep the parts that worked, you give credit, and the original author sees that someone found their work useful.

11. Comments

Every entity detail page has a comment section. You can discuss a prompt, explain a persona, or suggest an improvement for a constraint. Comments themselves can be voted on, and the list can be sorted by newest, oldest, highest rated, or lowest rated.

Comments are public: they appear to anyone who can view the entity. Signed-out visitors see the conversation but cannot join it.

When someone comments on one of your entities, you get a notification so you can follow up. See the next section.

12. Notifications

Prompty.tools keeps you informed about activity on your content. The Notifications entry in the dashboard sidebar shows an unread count as a badge, and the full list lives at /dashboard/notifications.

What you get notified about

• Comments on any entity you own: prompts, personas, tones, outputs, constraints, tone collections, constraint collections, and libraries. The notification includes a short preview of the comment.
• Remixes of your prompts. When another user remixes one of your prompts, you see who did it and the title of their new prompt.

Notifications are private to you. Nobody else can see what you have been notified about, and the sender is not told whether you read their comment or not.

Filtering and actions

• All / Unread tabs switch between the full history and only the items you have not yet read. Each tab shows its own count in the header.
• Clicking a notification takes you to the source (the commented entity, the remixed prompt) and marks it as read automatically.
• Dismiss a notification with the trash icon to remove it from the list permanently.
• Mark all as read clears the unread badge in one click without deleting any items.
• Pagination works the same as any other dashboard list, with selectable page sizes.

The sidebar badge caps its display at "9+" once you have more than nine unread items.

13. Collections

Collections let you group tones together, or constraints together, under a single name. Instead of selecting the same five tones every time you build a prompt, create a collection once and toggle the whole set with one click.

Collections are not building blocks. They never appear in the compiled prompt. They are an organizational tool that sits on top of existing tones and constraints.

Tone collections and constraint collections are separate

A collection holds either tones or constraints, never both. You manage tone collections from the Tones dashboard page and constraint collections from the Constraints dashboard page. Each side has its own list, its own quota, and its own favorites.

This split is intentional: tones and constraints play different roles in a prompt, and mixing them in one collection encouraged grouping that did not match how the builder applies them.

Properties

• Name: up to 80 characters. Names are case-insensitively unique per user, per collection type. You cannot create two tone collections called "Friendly" under your account, but a tone collection and a constraint collection can share a name.
• Description: up to 300 characters. Optional.
• Tags: up to 10, same rules as other entities.
• Visibility: public or private, same as other entities. Private collections count toward your tier limit.

Where collections live

Collections appear as a fourth tab on the Tones and Constraints dashboard pages. Inside that tab you can create, edit, delete, and browse collections of the matching type. Each collection has its own detail page where you can see its items, vote, favorite, comment, share, and embed.

Public collections also have dedicated pages outside the dashboard, accessible without an account:

• Tone collections are reachable at https://prompty.tools/tones/collections/{id}
• Constraint collections are reachable at https://prompty.tools/constraints/collections/{id}

These pages display the collection name, description, tags, the full list of tones or constraints it contains, the net vote score, the favorites count, and any public comments. Signed-out visitors can browse the content and read comments, but cannot vote, favorite, or comment without signing in. If you visit one of these URLs while already signed in, you are redirected automatically to the dashboard detail page for that collection.

The Share button on a collection's detail page copies the correct type-specific public URL to your clipboard. Note that the URL format is type-specific: a tone collection link and a constraint collection link use different paths, so sharing the wrong one will not work.

Collections in the builder

The Tone and Constraints sections of the prompt builder include a Collections filter alongside Custom, Public, Favourites, and All. Switching to the Collections filter shows your available collections as pills, scoped to the section you are in - you will not see constraint collections inside the Tone section, and vice versa.

• Toggle a collection to select or deselect all its items at once.
• Expand a collection to see individual items and toggle them one by one.
• Filter collections with a secondary pill bar (All, Mine, Public, Favourites) to narrow the list.
• Save as collection: when you have items selected, a "Save as collection" link lets you save your current selection as a new private collection of the matching type.

Social features

Collections support the same social features as every other entity: voting, favorites, comments, sharing by URL, and embedding as a read-only iframe. Public collections appear in the community's Public tab on the Tones and Constraints pages, in their respective lists. See Sharing and Embeds for the full details on how those features work.

Limits

Free accounts can create up to 5 private tone collections and up to 5 private constraint collections. The two quotas are tracked separately. Favorites are also tracked separately for each collection type. Pro removes both caps. Public collections do not count toward any cap. See Tiers & Access for all limits.

API

Tone collections and constraint collections are fully supported in the V1 API as separate resources, including creating, updating, deleting, managing items, voting, and favoriting. See the API documentation for the full reference.

14. Libraries

A library is a curated group of prompts. Where collections organize tones or constraints, libraries organize prompts themselves. Use a library to gather every prompt you wrote for customer support, every prompt your team uses for code review, or every prompt you want to keep on a single shareable page.

Libraries are not building blocks. They never appear in compiled prompts. They sit on top of your existing prompts and reference them.

Properties

• Name: up to 80 characters. Names are case-insensitively unique per user. You cannot have two libraries named "Code Review" on the same account.
• Description: up to 300 characters. Optional.
• Tags: up to 10, same rules as other entities.
• Visibility: public or private. Private libraries are a Pro-only feature; on Free, every library you create is public. See Tiers & Access.

Adding prompts to a library

Every prompt detail page has an Add to library button. Clicking it opens a modal where you can:

• Search and sort your existing libraries.
• Toggle a library on or off to add or remove the prompt from it.
• Create a new library on the spot, without leaving the modal.

A prompt can live in any number of libraries simultaneously. A library can hold any number of prompts. There is no separate "library version"; libraries always reflect the current set of memberships.

Where libraries live

Libraries have their own dashboard page at /dashboard/libraries with the standard Mine / Public / Favorites tabs. Each library has a detail page that lists its prompts, comments, votes, favorites, and metadata. Public libraries are also reachable by URL outside the dashboard at /libraries/{id}, the same way public prompts are.

Cross-references

A prompt's detail page shows the public libraries it belongs to, so visitors who like one prompt can discover the rest of the set. A user's profile page lists their public libraries alongside their public prompts.

Social features

Libraries support the same social features as every other entity: voting, favorites, comments, sharing by URL, and embedding as a read-only iframe on external sites. See Sharing and Embeds.

Limits

Free accounts can create unlimited public libraries but zero private ones. Pro raises this to unlimited private libraries. See Tiers & Access for the table.

API

Libraries are fully supported in the V1 API, including creating, updating, deleting, listing the prompts in a library, voting, and favoriting. See the API documentation for the full reference.

15. Queues

A queue is an ordered list of prompts designed to be worked through programmatically by a headless agent. Where the rest of your library is built for human use — browsing, copying, sharing, and community interaction — queues are an automation tool: you fill them with prompts, hand an agent the Queue ID and an API key, and let it claim and process each item in sequence.

Queues are always private. They have no public visibility toggle, no community browsing page, no voting, no favorites, and no comments. They are instruments for your own automation pipelines, not shareable community resources.

Properties

• Name: up to 80 characters. Required.
• Description: up to 300 characters. Optional.
• Queue ID: a UUID assigned when the queue is created. Copy it from the queue detail page and pass it to your agent's configuration.

Queue item statuses

Each item in a queue is a prompt paired with a status. The four statuses are:

Status	Meaning
Pending	Waiting to be claimed by an agent
In progress	Claimed by an agent; work is underway
Succeeded	The agent marked the item as completed successfully
Failed	The agent marked the item as failed

An agent claims the next pending item via the API and it immediately moves to In progress — there is no separate acknowledgement step. The agent then updates it to Succeeded or Failed when it finishes. You cannot manually move items between states or re-queue a failed item from the dashboard. Only Pending items can be manually removed.

The pending-items cap applies to Pending status only. Items in any other state do not count toward the cap. As an agent works through items and advances them to terminal states, the pending slots free up for new additions.

Adding prompts to a queue

There are two ways to add a prompt to a queue:

1. From a prompt's detail page: click Add to queue. A modal lists your queues with their current pending counts. Pick the target queue. You can also create a new queue from that modal without leaving the page.
2. From inside the queue: open the queue's detail page and click Add prompt. A modal lets you search your own prompts or public community prompts by name and add them directly.

In both cases, the prompt is added to the back of the queue with Pending status.

Working with queues in the dashboard

The Queues page at /dashboard/queues lists all your queues with their names, descriptions, and pending item counts. From here you can search, create, and delete queues.

On a queue's detail page:

• The name and description are editable at any time.
• The Queue ID is displayed with a one-click copy button — paste it into your agent's configuration.
• The items list shows all items with their status badges. Use the filter buttons at the top to narrow the view to a specific status: Pending, In progress, Succeeded, or Failed.
• Pending items can be individually removed with the trash icon. Items in other states cannot be removed.
• Pagination and page-size controls work the same as other dashboard lists.

Deleting a queue permanently removes it and all its items. There is no undo.

How agents use queues

The queue detail page shows the dequeue endpoint your agent should call. The call requires an API key; your tier's daily API request limit applies. See the API documentation for the full endpoint specification and for how an agent reports item outcomes.

A typical automated pipeline looks like this:

1. Populate the queue with prompts through the dashboard or via the API.
2. Configure your agent with the Queue ID and an API key.
3. The agent calls the dequeue endpoint in a loop, claiming and processing items one at a time.
4. Each item is marked Succeeded or Failed by the agent as it finishes.
5. Monitor progress from the queue detail page without interrupting the agent.

When the queue is empty, the dequeue endpoint returns a response indicating no items are available — the agent can use this as its stop signal.

Limits

Limit	Free	Pro
Queues	1	Unlimited
Pending items per queue	3	Unlimited

On the Free tier, the single-queue limit means you must delete a queue before creating another. Pro removes both caps entirely.

16. Sharing

Every detail page has a Copy button that sends the compiled prompt, persona text, or block content straight to your clipboard. For public items, the page URL itself is shareable: paste it anywhere and the recipient can view the item, vote on it, and favorite it.

Signed-out visitors can view public pages, read comments, and follow links, but they cannot vote, favorite, or comment. When you share a link, expect a signed-out recipient to hit a sign-in prompt when they try to interact.

17. Embeds

Every public entity in Prompty.tools (prompts, personas, tones, outputs, constraints, collections, and libraries) can be embedded on external websites as a compact, read-only card inside an iframe.

Getting the embed code

On any detail page you own, an Embed button copies an iframe snippet to your clipboard. Paste it into any HTML page, blog post, or CMS that supports raw HTML. The button is disabled for private items because embeds only work for public content.

What the embed shows

The embedded card displays the entity type badge, title, description or content preview, tags, voting score, favorite count, comment count, author, and date. Prompts also show a preview of the compiled text. Tone and constraint collections show their grouped items. Libraries show a preview of the prompts they contain. A link at the bottom takes the viewer to the full detail page on Prompty.tools.

Auto-resizing

The embed snippet includes a script that automatically adjusts the iframe height to fit the card content. Include both the <script> tag and the <iframe> tag from the snippet for this to work.

Privacy and indexing

Only public items can be embedded. If the author makes an item private or deletes it after the embed code has been shared, the embed shows a "Content unavailable" fallback instead. Embed pages are not indexed by search engines.

18. Reporting content

If you find a public building block or prompt that violates community standards — spam, harmful content, or anything that does not belong — you can flag it for moderator review directly from its detail page.

How to report

1. Open the detail page of the prompt, persona, tone, output format, or constraint you want to flag.
2. Click the Report button (flag icon). It appears near the action controls on the detail page.
3. Describe the issue in the text field. Up to 500 characters. The reason must not be empty.
4. Click Submit report.

A moderator reviews every report and decides whether to act on it. You will not receive a direct reply, but Prompty.tools confirms the submission with a notification.

What to report

Report content that is genuinely problematic: spam, abusive text, content that violates the terms of service. The report mechanism is not a vote-down substitute — if you disagree with a building block's approach or find it unhelpful, use the downvote instead. Misusing the report system wastes moderator time.

What happens next

Moderators can dismiss a report (no action taken) or remove the offending content. The process is internal. Neither the reporter nor the author of the flagged content is notified of the specific outcome.

Sign-in required

You must be signed in to submit a report. Signed-out visitors cannot access the Report button.

19. Reputation and the community leaderboard

Reputation is a numeric score that reflects how much the community has recognized your public contributions. It increases when other users vote on or favorite your public building blocks and prompts. It is displayed on your public profile page and on the Settings page alongside your content counts.

How Reputation is calculated

Every upvote and every favorite received on a public entity contributes to your Reputation score. Not all entity types carry equal weight — prompts contribute more per interaction than individual tones or constraints do, because prompts represent more complete, higher-effort work. Favorites count more per event than upvotes. Downvotes reduce the score.

The exact weights are an implementation detail and may be adjusted. The principle is stable: create high-quality, useful public building blocks, and Reputation follows.

Where Reputation appears

• Your public profile — visible to anyone who visits your profile at /users/{username}. The score is shown with two delta values: how much it increased in the last month and in the last week. Alongside the score you will find a per-type breakdown of your content counts (prompts, personas, tones, outputs, constraints), each color-coded to match the builder.
• Other users' profiles — shown in the same position, so you can compare at a glance.
• Settings page — displayed alongside your content counts in the "Your Data" section.

Reputation is always calculated from your public content only. Private items you create do not contribute to your score, and nobody can see your private items' interaction data.

The community leaderboard

The Users page at prompty.tools/users ranks all users by their Reputation score. It is public — no sign-in required to view it.

• Three time windows: All time, Last month, and Last week. Switch between them using the tabs at the top of the page.
• Pagination: 25 users per page. Use the Previous and Next links to navigate.
• Profile links: every entry on the leaderboard links to that user's public profile.

The leaderboard is also reachable via the Users link in the dashboard sidebar, placed alongside Libraries.

Activity chart

Every user profile page also displays a monthly activity chart: a stacked bar graph covering the last 12 calendar months of content creation. Each bar is broken down by type — prompts, personas, tones, outputs, and constraints — using the same color coding as the builder. Hovering a bar segment shows a tooltip with exact counts for that month.

The chart reflects public and private content alike (for your own profile) and public content only (for other users' profiles). It gives a quick read on someone's pace of work and what kinds of building blocks they focus on.

Reputation is a signal, not a goal

The leaderboard reflects community engagement, not absolute quality. A user with fewer but highly-voted prompts can rank above a user with a large volume of low-engagement work. Focus on publishing building blocks that other users genuinely find useful, and the score takes care of itself.

20. AI suggestion learning

When you type a task in the prompt builder, Prompty.tools can suggest building blocks to go with it — personas, tones, output formats, and constraints that fit the task text. Over time, the suggester adapts to your choices.

How it works

The system learns from two explicit signals only. Keeping a suggestion teaches it nothing, because suggestions are applied automatically and leaving them selected is simply the default state.

• Items you remove after they are suggested are excluded from future runs for similar tasks, so the same unhelpful suggestions stop appearing.
• Items you add yourself that were not suggested are boosted for similar tasks, making them more likely to surface next time.

The learning is personal: it applies only to your account and does not influence what other users see.

Clearing your learning data

If you want suggestions to start fresh — for example, if your workflow has shifted and past rejections are now filtering out useful suggestions — go to Settings → Clear AI suggestion learning and confirm. The action removes all stored signal data immediately. Suggestions will return to their unbiased defaults for your next session.

This action is permanent

Clearing suggestion learning cannot be undone. Past feedback is deleted, not archived. If you later find the unbiased suggestions unhelpful, you will need to use the builder a few times to rebuild the signal.