Prompt2Love gives you one personal library for two kinds of content. Prompts are full, ready-to-use prompt texts you fire off in a single go. Skills are reusable instruction templates paired with a 'when to use' description that tells an AI tool how and when to apply them. Both live side by side in your dashboard, both support categories, tags, favorites, {{variables}} and forking, and both track when you last used them - so the stuff you actually rely on never gets lost in a chat history again.
Prompts vs. Skills
Two content types for two jobs: complete prompts you fire off, and reusable instruction templates with a 'when to use' description.
A prompt holds a full prompt text in its content field and carries a workflow status, output capture and community engagement metrics, so it is built for complete AI requests - for example 'Write a social post about {{topic}}'. A skill holds its instructions plus an optional 'When to Use' trigger description, a target model and a source URL, but no status and no output capture, so it is built for repeatable instruction patterns - for example 'Act as {{role}} and respond in {{tone}}'. Both types can be forked, favorited, tagged, categorized and filled with variables. The dashboard shows prompts and skills in separate tabbed sections so the two never get mixed up.
- •Prompt fields: title (max 200) and content (max 50,000) required; optional description (max 500), tags, language, output type, AI tool, status
- •Skill fields: title (max 200) and instructions (max 50,000) required; optional description (max 500), 'When to Use' trigger (max 500), model, tags, language, source URL
- •Skills have no status workflow, no output capture and no per-record community vote, view or fork metrics
- •Both support {{variable}} syntax, a favorite toggle, last-used tracking, soft delete and cover/display images
- •Free tier: 30 prompts and 10 skills
- •Prompts and skills sit in separate tabbed sections of the dashboard
Creating and Editing
Create or edit prompts and skills in a two-tab dialog for Content and Variables; edits save only what changed and duplicates are blocked.
The create and edit dialog has two tabs: 'Inhalt' (Content), where you write the prompt text or skill instructions, and 'Variablen' (Variables), where you manage placeholders. When you save, only the fields you actually changed are sent. A content fingerprint detects duplicates and blocks creating two items with identical content - unless you are forking something from the public community, where identical copies are allowed. If you publish a prompt without a description, the system auto-fills a meta description from the first 155 characters of your cleaned content. An unsaved-changes warning catches you before you close the dialog by accident.
- •Two-tab dialog: 'Inhalt' (Content) and 'Variablen' (Variables)
- •Duplicate detection via content fingerprint; skipped when you fork from the community
- •Publishing a prompt without a description auto-fills one from the first 155 characters of cleaned content
- •Variables are auto-detected from {{...}} patterns as you type
- •Unsaved-changes confirmation before you discard
Status: The Prompt Workflow State
Each prompt carries a five-stage workflow status from 'new' to 'completed', with no forced order between stages.
Every prompt has a status that defaults to 'neu' (new). The workflow is deliberately flexible: you can set any status at any time and jump straight from 'neu' to 'abgeschlossen' (completed) if you want - nothing forces a particular order, and only the owner can change it. The most natural place to manage status is the Kanban view, which turns the five statuses into five fixed, color-coded columns you drag cards between. Status is there to organize your work, not to police it. Skills have no status field.
- •Five values: 'neu' (new), 'offen' (open), 'in_bearbeitung' (in progress), 'review', 'abgeschlossen' (completed)
- •Default on creation: 'neu'
- •Owner-only; no enforced transitions - any jump is allowed
- •Surfaced as the five Kanban columns (Neu / Offen / In Bearbeitung / Review / Abgeschlossen); changed by dragging cards there
- •Skills have no status field
Categories
Nested folders with unlimited depth; a prompt or skill can live in several categories at once.
Categories organize your library like nested folders, with unlimited depth, expand and collapse in the sidebar, and item-count badges that include all child categories. Each category has a name, an icon, a parent and a position. An item is not locked into a single folder - it links to categories many-to-many, so one prompt can sit in several categories at once. You set categories when you create or update an item, and you reorder or re-parent them by dragging in the sidebar. In the Kanban view you can also drag prompts between category columns.
- •Many-to-many: one prompt or skill can belong to several categories
- •Category fields: name, icon, parent, position; system categories are flagged as seed data
- •Built-in filters 'Favorites' and 'Uncategorized'
- •Public community categories: SEO, Leads, Texting, Webdesign, Vibe Coding, Marketing, Sales, Content, Other
- •Drag to reorder or re-parent in the sidebar; drag between category columns in Kanban
Favorites
A simple bookmark toggle on each of your own prompts and skills.
Favorites are a personal bookmark on an item. The toggle is idempotent - tap it once to favorite, tap again to un-favorite - and you can only favorite your own items. Your favorites feed the sidebar 'Saved' filter and the 'favorites' sort option, so the prompts and skills you reach for most are always one click away. This is your own private bookmark, not a public like count - it is separate from community likes and saves on posts.
- •Idempotent toggle: tap to favorite, tap again to remove
- •Only on your own items
- •Feeds the sidebar 'Saved' filter and the 'favorites' sort
- •Private bookmark, not a public like count - separate from community likes and saves
Forking: Copy to Your Library
Copy any public community prompt or skill into your own library as a private copy that doesn't count against your quota.
Forking takes a public community prompt or skill and drops a private copy into your library, owned by you. The fork remembers where it came from and keeps the original author's attribution, and it inherits the title, description, content or instructions, tags, AI tool, language and type. Best of all, forking bypasses both your quota limit and the duplicate check, so you can collect identical content freely. The link is one-directional: your fork knows its source, but editing the fork never writes back, and the source never changes your copy. Forking from the community also bumps the original post's fork count and notifies its author.
- •Creates a private copy owned by you; records the source item and original author
- •Bypasses both your quota limit and the duplicate check
- •Sets the source URL to /community/post/{slug} so re-shares auto-detect their source
- •One-directional link: no edit sync back to the source
- •Forking a community post increments its fork count and notifies the author
Version History
Every content edit auto-saves a snapshot of the previous version; browsing and managing versions is a Pro feature.
Whenever a prompt's content changes, the system quietly snapshots the previous version in the background, with an auto-incremented version number that stays correct even when you make several edits in quick succession. Each snapshot keeps the old title, content, result, model and timestamp, and you can also create a manual snapshot. Versions are append-only and immutable: you can list them newest-first and delete a version, but there is no one-click restore, so rolling back to an older version is done by hand. Creating versions is a Pro-tier feature and is not available on the free tier.
- •Automatic snapshot on every content edit, taken in the background
- •Each snapshot stores version number, title, content, result, model and timestamp
- •Version numbers stay correct even across rapid consecutive edits
- •Pro feature: the free tier cannot create versions
- •Append-only: list newest-first and delete a version, but no restore or revert - roll back by hand
Permissions and Access Control
Every read and write runs through centralized permission checks based on ownership, team membership, role and visibility.
Access to your prompts and skills runs through a single, centralized set of permission checks rather than scattered ad-hoc rules, so the same logic governs every read and write. Those checks weigh ownership, team membership, role and whether an item is public. As the owner you can edit, delete, favorite and reorder your items; deleting is a soft delete, so an item is hidden rather than wiped, and you can reorder items by position within a category. The result is consistent, predictable access control across the whole library.
- •Centralized permission checks, never inlined per route
- •Checks weigh ownership, team membership, role and the public/private flag
- •Owner can edit, delete, favorite and reorder; team editors can access team items
- •Soft delete hides an item instead of wiping it
- •Reorder items by position within a category
