This directory contains specialized AI prompts for working with the ObjectStack specification repository. Each prompt defines a specific role, context, and responsibilities to help AI assistants provide focused, high-quality assistance.
These prompts help define the core ObjectStack protocols:
Location: packages/spec/src/data/
Focus: ObjectQL - Data structure, validation, permissions, workflows
Key Responsibilities:
- Field definitions (23+ types)
- Object schemas
- Validation rules
- Permission systems
- Workflow automation
- Flow builders
- Query AST
- Trigger contexts
Use When: Defining data models, business logic, or data access patterns
Location: packages/spec/src/ui/
Focus: ObjectUI - Server-Driven UI definitions
Key Responsibilities:
- View protocols (List, Form, Calendar, Kanban, Gantt)
- App navigation structures
- Dashboard layouts and widgets
- Report definitions
- Action buttons
- Page layouts
- Theme configurations
- Widget contracts
Use When: Defining UI structures, views, or user interactions
Location: packages/spec/src/system/
Focus: ObjectOS - Runtime environment and platform capabilities
Key Responsibilities:
- Manifest (packaging)
- Plugin lifecycle
- Driver interface
- Identity & authentication
- Role-based access control
- API contracts
- Webhooks
- Translations (i18n)
- Multi-tenancy
Use When: Defining system-level features, plugins, or platform capabilities
Location: packages/spec/src/ai/
Focus: AI agent integration
Key Responsibilities:
- Agent definitions
- Tool integrations
- Knowledge bases (RAG)
- Conversation management
- Prompt templates
- AI workflows
- Model configurations
Use When: Integrating AI capabilities, defining agents, or building RAG systems
Location: packages/spec/src/api/
Focus: API contracts and standardization
Key Responsibilities:
- Response envelopes
- Request schemas
- API contracts
- Error codes
- REST endpoints
- GraphQL schemas
- Webhook payloads
- Batch operations
Use When: Defining API structures, endpoints, or integration contracts
These prompts help with specific development tasks:
Focus: Comprehensive test coverage for all protocols
Key Responsibilities:
- Schema validation tests
- Type inference tests
- Edge case testing
- Integration tests
- Regression tests
- Performance tests
- Documentation tests
Use When: Writing tests, improving coverage, or validating schemas
Focus: Clear, comprehensive documentation
Key Responsibilities:
- TSDoc comments
- Concept documentation
- API reference docs
- Tutorials and guides
- Migration guides
- Example documentation
Use When: Writing docs, adding comments, or creating tutorials
Focus: Realistic, runnable examples
Key Responsibilities:
- Full application examples (CRM, Todo, etc.)
- Quick start examples
- Feature-specific examples
- Real-world scenarios
- Advanced examples
- Example documentation
Use When: Creating examples, demos, or reference implementations
These prompts are compatible with multiple AI coding assistants:
- Claude Code (Recommended): Automatically loaded from the repository context
- GitHub Copilot: Automatically loaded when working in the repository
- Cursor: Can reference these prompts for context-aware assistance
They provide context-aware assistance based on:
- The file you're editing
- The protocol layer you're working on
- The task you're performing
- Choose the Right Prompt: Select the prompt that matches your task
- Provide Context: Share relevant code, requirements, or issues
- Be Specific: Clear instructions get better results
Each prompt includes interaction commands for common tasks:
Data Protocol:
- "Create Field Protocol" → Implement field definitions
- "Create Object Protocol" → Implement object schemas
- "Create Validation Rules" → Implement validation engine
UI Protocol:
- "Create View Protocol" → Implement view definitions
- "Create Dashboard Protocol" → Implement dashboard layouts
- "Create Theme System" → Implement theming
System Protocol:
- "Create Manifest Protocol" → Implement packaging
- "Create Plugin System" → Implement plugin lifecycle
- "Create Driver Interface" → Implement database drivers
AI Protocol:
- "Create Agent Protocol" → Implement AI agents
- "Create Knowledge Base" → Implement RAG system
Testing:
- "Write tests for X" → Create comprehensive tests
- "Add edge case tests" → Add boundary tests
Documentation:
- "Document X schema" → Write complete docs
- "Add TSDoc comments" → Add inline documentation
Examples:
- "Create CRM example" → Build full CRM app
- "Add lookup example" → Show relationship patterns
All prompts enforce these standards:
- Configuration Keys (TS properties):
camelCase- Example:
maxLength,referenceFilters,defaultValue
- Example:
- Machine Names (data values):
snake_case- Example:
first_name,project_task,account_id
- Example:
// 1. Define Zod schema
export const MySchema = z.object({
field: z.string().describe('Purpose'),
});
// 2. Infer TypeScript type
export type My = z.infer<typeof MySchema>;- Every field must have
.describe()annotation - Complex schemas need JSDoc comments
- Include examples in tests and docs
- 80%+ code coverage target
- Test valid inputs, invalid inputs, edge cases
- Integration tests for connected schemas
-
Design Phase
- Use Protocol Architect prompts to define schemas
- Reference existing protocols for consistency
-
Implementation Phase
- Use architect prompts to implement Zod schemas
- Follow naming conventions strictly
-
Testing Phase
- Use Testing Engineer prompt
- Achieve 80%+ coverage
-
Documentation Phase
- Use Documentation Writer prompt
- Add TSDoc and concept docs
-
Example Phase
- Use Example Creator prompt
- Create runnable examples
- Follows naming conventions (camelCase/snake_case)
- Zod schema with
.describe()annotations - Type inferred from Zod
- Unit tests with 80%+ coverage
- TSDoc comments
- Documentation updated
- Example created/updated (if major feature)
- Zod First: Always start with Zod schema, derive types
- Strict Types: Never use
any, use proper unions - No Business Logic: Only definitions, no implementations
- Extensibility: Design for plugin additions
- Backwards Compatibility: Never break existing APIs
When designing protocols, benchmark against:
- Salesforce: Object/Field model, Lightning components
- ServiceNow: Service catalog, CMDB
- Kubernetes: Manifest structure, CRDs
- OpenAPI: API contract definitions
- Data as Code: All metadata is versioned
- Idempotency: Same input = same output
- Immutable Infrastructure: No runtime schema mutations
- Convention over Configuration: Sensible defaults
- ARCHITECTURE.md - System architecture
- DEVELOPMENT_ROADMAP.md - Development plan
- PRIORITIES.md - Priority matrix
- README.md - Project overview
- CRM Example - Complete application
- Todo Example - Quick start
- Feature Examples - Specific features
When adding new prompts:
- Follow Template: Use existing prompts as template
- Define Scope: Clear role and responsibilities
- Include Examples: Show expected patterns
- Add Commands: List interaction shortcuts
- Update Index: Add to this README
# 🎯 [Role Name]
**Role:** You are the **[Title]** for ObjectStack.
**Context:** [What you're defining]
**Location:** [Directory or scope]
## Mission
[One-paragraph mission statement]
## Core Responsibilities
### 1. [Responsibility Name]
[Details with code examples]
## Coding Standards
[Standards specific to this role]
## Interaction Commands
[Quick commands for common tasks]
## Best Practices
[Tips and guidelines]
## Reference Examples
[Links to relevant examples]- Discord: Join our community
- GitHub Issues: Report issues
- Documentation: Read the docs
These prompts are part of the ObjectStack Protocol specification.
License: Apache 2.0 © ObjectStack
Last Updated: 2026-01-21
Version: 1.0.0
Maintainer: ObjectStack Team