Variables allow you to create dynamic prompts that can be customized each time you execute them. Use the syntax {{variableName}} in your prompt text. When you execute the prompt, you'll be prompted to fill in values for each variable.

For example, if your prompt contains {{topic}} , you'll see an input field for topic when executing. Variables make prompts reusable across different scenarios without needing to rewrite them.

Contexts are additional information that gets prepended to your prompt when executing. They're useful for providing background information, instructions, or data without cluttering your main prompt.

When you copy a prompt, you can choose which linked contexts to include. Contexts are formatted with clear headers to help AI models understand the structure. You can link multiple contexts to a single prompt, and the same context can be used across many prompts for consistency.

PromptCrew uses a hierarchical permission system:

• Private: Only you can see and use the prompt, context, or collection.
• Company: All members of your company can access it.
• Team: Only members of your specific team can access it.
• Group: Only members of selected groups can access it. Groups can contain multiple teams.
• Custom: Fine-grained control where you select specific users, teams, or groups and assign different permission levels (view-only, edit, owner).

You can also set permission levels such as view-only (can see but not use), edit (can modify), and owner (full control including deletion and permission management).

When you grant access to users, teams, or groups, you can assign one of three permission levels:

• View-Only: Users can see the title, description, and basic information about the prompt, context, or collection, but cannot view full content, execute prompts, or make changes.
• Edit: Users can view full content, execute prompts, copy prompts, and make modifications, but cannot delete items or change permissions.
• Owner (Co-Owner): Users have full control, including editing, deleting, and managing permissions. Only creators and owners can publish to the community or change community status.

The creator of an item always has owner-level access and cannot be removed. Permission levels apply to prompts, contexts, and collections.

Custom permissions give you fine-grained control over who can access your prompts, contexts, and collections. You can:

• Control access: Grant access to specific users, teams, or groups instead of sharing with an entire company or team.
• Assign roles for each entry: View-Only, Edit, or Owner.
• Mix access levels: Some users can have edit access while others are view-only.
• Use collections: Collections control visibility of titles and descriptions, while individual item permissions still control full access.
• Permission inheritance: When you share a collection, users need both collection access and individual item permissions to fully use the content.

Custom permissions are ideal when you need to share with specific people outside your immediate team or when you want different access levels for different users.

The Community Library is a public repository where all PromptCrew users can discover and use prompts shared by the community. When you publish a prompt to the community, it becomes publicly available to all users.

Important: All prompt information, including content, variables, description, and any associated context documents, will be made publicly available. We strongly recommend against sharing sensitive company information, proprietary data, or confidential content in community prompts. Community prompts cannot be added to collections.

There are several ways to share:

• Shareable links: Use the \"Share\" feature to generate a unique link. Anyone with the link can access the item (if they have permission).
• Permission settings: Set permissions to company, team, or group level to automatically share with those users.
• Community Library: Publish prompts to the community to make them available to all PromptCrew users.
• Collections: Share entire collections with prompts and contexts together.

When sharing via link, users still need appropriate permissions to view full details. Those without permission will see an error message.

Yes. The Prompt Builder includes a built-in test modal. Click Execute Prompt in the Build tab to test your prompt with different AI models.

You can test with multiple models simultaneously to compare responses. This helps you refine your prompt before making it available to others. You can also test prompts from the Prompt Library after they're published.

Currently, you can upload TXT (plain text), DOCX (Microsoft Word), and PDF files. The system automatically extracts and parses the content from these files.

For best results, ensure your documents are well-formatted and contain clear, readable text. File uploads are processed server-side, and the extracted content is stored in your Context Repository for reuse across multiple prompts.

Yes. You can edit any prompt you created or have edit permissions for. When you save changes, a new version is created automatically, and you can view the complete version history to see all changes over time.

Version history shows who made changes and when, and you can restore previous versions if needed. This helps you maintain a record of how your prompts have evolved and revert to earlier versions when necessary.

Collections help you organize related prompts and contexts together:

• Adding items: When creating or editing prompts and contexts, you can assign them to a collection.
• Visibility: Anyone with access to a collection can see the titles and descriptions of items in it. To view full details and use prompts/contexts, users need individual permissions for each item.
• Organization: Collections are perfect for grouping work by project, workflow, or use case.
• Sharing: Share entire collections with teams or groups, but remember that individual item permissions still apply.

Community prompts cannot be added to collections.

Drafts are prompts you're still working on. They're saved but not yet published, so they're only visible to you. Published prompts are finalized and available to others based on their permission settings.

You can save prompts as drafts at any time during the creation process using Save Draft, then publish them when ready. Drafts let you iterate without making prompts visible to your team until they're complete. You can access all drafts from the Drafts section.

Cloning creates a copy of an existing prompt that you can edit independently. When you clone a prompt, you get a copy with all the original content, variables, contexts, and settings.

The cloned prompt becomes your own, and you can modify it without affecting the original. This is useful for adapting prompts to new use cases or creating variations. Cloned prompts start as drafts so you can refine them before publishing.

You can mark prompts as favorites by clicking the star icon on any prompt card. Favorites are saved to your personal list for quick access.

The Recent view shows prompts you've recently viewed or executed, making it easy to return to prompts you're actively working with. Both features help you quickly access frequently used prompts without searching your entire library.

The search feature allows you to search across prompts, contexts, and collections. You can search by title, description, content, tags, or category. Results only include items you have permission to access.

Use the main search page for comprehensive searches, or use the search bar in the Prompt Library, Context Repository, or Collections pages for filtered searches within those sections.

Use the Copy Prompt feature in the prompt dialog. This opens a modal where you can select which linked contexts to include.

The copied text is formatted in an LLM-friendly way with clear headers for each context, followed by the main prompt. This makes it easy to paste into ChatGPT, Claude, or other AI interfaces and helps models distinguish between context information and the main prompt.

If you delete a context, it will be removed from all prompts that were linked to it. The prompts themselves remain, but they'll no longer include that context when executed.

Make sure to review which prompts use a context before deleting it, as this action cannot be undone. You can see usage information for contexts in the Context Repository to help you make informed decisions.

Categories are predefined labels (like \"Engineering\", \"Marketing\", \"Sales\") that help group related prompts and contexts. Tags are custom labels you can add for more flexible organization.

Both help you filter and find content quickly. You can filter by category in the Prompt Library and Context Repository, and search by tags. Categories are typically managed by admins, while tags can be added by anyone creating or editing content.

Different models have different strengths. For example:

• GPT-4 is excellent for complex reasoning and multi-step tasks.
• Claude 3 is great for long-form content and analysis.
• GPT-3.5 Turbo is faster and more cost-effective for simpler tasks.

Use the test modal in the Prompt Builder to try your prompt with multiple models and see which gives the best results for your use case. You can also set a default model for a prompt, and admins can configure which models are available to your team.

Admin users have access to several management features:

• Users & Roles: Manage team members and assign roles.
• Account Settings: Configure company-wide settings, categories, and preferences.
• Analytics: View usage statistics, popular prompts, and team activity.
• Billing: Manage subscriptions, view usage, and handle billing information.

Admin access is typically granted to company owners and designated administrators.

Currently, you can copy prompts (with or without contexts) using the Copy Prompt feature, which formats everything in an LLM-friendly way. This is the quickest way to export individual prompts for use in other tools.

Full export functionality for bulk backups or migration is planned for a future release. For now, the copy feature provides the most practical way to export prompts with their contexts.