Prompt Library
Create, manage, and version prompts for consistent AI interactions across your organization.
The Prompt Library is a powerful feature that lets you create, manage, and version prompts for AI agents. It ensures consistency across your organization and allows you to optimize prompts over time.
Why Use a Prompt Library?
Without a Library
- Each team member writes prompts differently
- No way to share effective prompts
- Can't track what worked and what didn't
- Prompts get lost in chat history
With Fabric's Prompt Library
- Consistency — Everyone uses the same optimized prompts
- Versioning — Track changes and rollback if needed
- Sharing — Share effective prompts across teams
- Analytics — See which prompts perform best
- Templates — Use variables for dynamic content
Prompt Types
System Prompts (Read-Only)
Platform-provided prompts optimized for specific tasks:
- PRD Generator — Creates Product Requirement Documents
- Technical Spec — Generates technical specifications
- Architecture Doc — Designs system architecture documents
- User Story — Creates user stories with acceptance criteria
- API Documentation — Documents API endpoints
These prompts are:
- Tested and optimized by Fabric
- Can't be edited directly
- Can be forked to create custom versions
User Prompts (Editable)
Custom prompts created by you or your organization:
- Full editing control
- Version history
- Organization-wide or personal scope
- Can be based on system prompts
Creating Prompts
Navigate to Prompt Library
Click Settings → Prompt Library in the sidebar.
Create New Prompt
Click Create Prompt and fill in:
- Name — Descriptive name (e.g., "Engineering PRD Template")
- Description — What the prompt is for
- Content — The actual prompt text
Add Template Variables
Use Handlebars syntax for dynamic content:
You are writing a PRD for {{projectName}}.
The target audience is {{audience}}.
Requirements:
{{#each requirements}}
- {{this}}
{{/each}}
{{#if includeTimeline}}
Include a timeline section with milestones.
{{/if}}Set Scope
Choose who can use this prompt:
- Personal — Only you
- Organization — Everyone in your org
- System — Platform-wide (admin only)
Save
Click Save to create the prompt. It's now available for use.
Template Variables
Template variables make prompts dynamic and reusable.
Syntax
{{variableName}} — Simple variable
{{#if condition}}...{{/if}} — Conditional block
{{#each items}}...{{/each}} — Loop over array
{{#unless condition}}...{{/unless}} — Negative conditionalBuilt-in Variables
Fabric automatically provides these variables:
| Variable | Description |
|---|---|
{{projectName}} | Current project name |
{{projectDescription}} | Project description |
{{organizationName}} | Organization name |
{{userName}} | Current user's name |
{{currentDate}} | Today's date |
{{documentType}} | Type of document being generated |
Custom Variables
Define your own variables when using prompts:
Create a {{documentType}} for {{featureName}}.
Target users: {{targetUsers}}
Priority: {{priority}}When using this prompt, you'll be asked to provide values for each variable.
Prompt Bindings
Bindings connect prompts to specific agents or features.
How Bindings Work
┌─────────────────────────────────────────────────────────┐
│ Prompt Resolution │
├─────────────────────────────────────────────────────────┤
│ │
│ 1. User-selected prompt (from dropdown) │
│ ↓ if not selected │
│ 2. User-bound prompt (personal default) │
│ ↓ if not found │
│ 3. Org-bound prompt (organization default) │
│ ↓ if not found │
│ 4. System-bound prompt (platform default) │
│ ↓ if not found │
│ 5. Fallback (hardcoded prompt) │
│ │
└─────────────────────────────────────────────────────────┘Creating a Binding
Select a Prompt
In the Prompt Library, click on a prompt to view details.
Click "Bind"
Click the Bind button to create a binding.
Choose Target
Select what to bind to:
- Agent — Use this prompt for a specific agent
- Document Type — Use for generating specific document types
- Feature — Use for specific features (e.g., chat, generation)
Set as Default
Toggle Set as Default if this should be the automatic choice.
Versioning
Every prompt maintains a version history.
Automatic Versioning
When you edit a prompt:
- A new version is created automatically
- Previous versions are preserved
- You can view the diff between versions
- Rollback is available anytime
Version History
v3 (current) — Added timeline section
v2 — Improved formatting
v1 — Initial versionRollback
To rollback to a previous version:
- Open the prompt
- Click Version History
- Select the version to restore
- Click Restore This Version
A new version is created (rollbacks don't delete history).
Organization Prompts
Sharing Prompts
Share prompts with your organization:
- Create or edit a prompt
- Set scope to Organization
- Save the prompt
All organization members can now:
- View the prompt
- Use the prompt
- Fork to create personal versions
Admin Controls
Organization admins can:
- Create organization-wide prompts
- Set default prompts for the organization
- Manage prompt access permissions
- View usage analytics
Best Practices
Writing Effective Prompts
Be Specific
❌ "Write a document about the feature"
✅ "Write a PRD for {{featureName}} that includes:
- Problem statement
- Proposed solution
- User stories
- Success metrics"Use Context
❌ "Create user stories"
✅ "Based on the PRD for {{projectName}}, create user stories
that follow the format:
As a [user type], I want [goal] so that [benefit]"Include Examples
"Format the output like this example:
## User Story 1
**As a** customer
**I want** to reset my password
**So that** I can regain access to my account
### Acceptance Criteria
- [ ] User receives reset email within 1 minute
- [ ] Reset link expires after 24 hours"Prompt Organization
Naming Convention
[Type] - [Purpose] - [Version/Variant]
Examples:
"PRD - Feature Request - Technical"
"Story - Bug Fix Template"
"Spec - API Documentation - REST"Categories
- Document generation prompts
- Code review prompts
- Analysis prompts
- Communication prompts
Testing Prompts
Before making a prompt organization-wide:
- Test with various inputs
- Check output quality
- Verify variable handling
- Get feedback from team members
- Iterate based on results
Analytics
Track prompt performance:
| Metric | Description |
|---|---|
| Usage count | How many times used |
| Satisfaction rate | User ratings |
| Edit rate | How often output is edited |
| Time to complete | Generation speed |
Access analytics in Settings → Prompt Library → Analytics.