Skip to content

adanomad/pronghorn

BranchesTags
 
 

Repository files navigation

Pronghorn (Alpha)

Build Software with AI-Powered Precision

A standards-first, agentic AI platform that transforms unstructured requirements into production-ready code with complete traceability. From idea to deployment, Pronghorn orchestrates multi-agent AI teams to design, build, and ship software autonomously.

License: MIT Built with Lovable Powered by Supabase

Live Application: https://pronghorn.red

Table of Contents

Overview

Pronghorn is an open-source AI-powered software development platform created by the Government of Alberta, Ministry of Technology and Innovation. It enables teams to:

  • Design → Transform ideas into structured requirements with AI decomposition
  • Build → Generate production code with autonomous AI coding agents
  • Ship → Deploy to cloud platforms with integrated CI/CD

The platform operates in multiple modes:

  1. Design Mode: Visual specification building with React Flow canvas
  2. Audit Mode: Multi-agent cross-comparison between project datasets
  3. Build Mode: Autonomous code generation with real-time monitoring
  4. Present Mode: AI-generated project presentations with blackboard reasoning

Core Features

🎯 AI-Powered Requirements

Transform unstructured ideas into hierarchical specifications:

  • Epics → Features → User Stories → Acceptance Criteria
  • AI decomposition via LLM providers (Gemini, Claude, Grok)
  • Automatic linking to organizational standards for complete traceability

📋 Global Standards Library

Reusable compliance templates across your organization:

  • User-customizable categories and hierarchical trees
  • Dynamic linking to all projects (updates propagate automatically)
  • Tech stack templates with associated standards

🎨 Visual Architecture Design

Interactive canvas for system design:

  • 24+ node types (WEB_COMPONENT, API_ROUTER, DATABASE, SCHEMA, TABLE, etc.)
  • Data-driven node types from database (add types without code changes)
  • Real-time collaboration with multi-user editing
  • Layer management for complex diagrams
  • Lasso selection for bulk operations

📊 AI Presentation Generator

Create professional Gamma-style presentations from project data:

  • Blackboard Memory Pattern: Agent sequentially reads project data, accumulates insights
  • 15+ Slide Layouts: Cover, bullets, stats grid, architecture, timeline, comparison, etc.
  • Multiple Themes: Professional Dark, Clean Light, Vibrant Gradient
  • Mode Options: Concise (10-15 slides) or Detailed (20-30 slides)
  • Cover Image Generation: AI-generated cover images with style selection (Photorealistic, Illustrated, Abstract, Cartoon, Whiteboard, Infographic)
  • PDF Export: High-quality PDF generation with slide notes (low/high resolution)
  • Real-time Editing: Edit slides, notes, layouts, and font scaling per slide
  • JSON/Markdown Export: Download for backup or import into external tools

🔍 Multi-Agent Audit System

Cognitive cross-comparison between project datasets:

  • Dataset Comparison: Compare Requirements ↔ Canvas, Canvas ↔ Repository, Standards ↔ Code
  • 1:1 and 1:Many Modes: Compare single pairs or iterate one dataset against another
  • Tesseract Evidence Grid: 3D visualization (X: elements, Y: iterations, Z: polarity scores)
  • Knowledge Graph: Visual concept mapping with force-directed layout
  • Multi-Perspective Agents: Security Analyst, Business Analyst, Developer, End User, Architect
  • Venn Diagram Output: Unique to D1, Aligned, Unique to D2 (Challenges)
  • Blackboard Collaboration: Agents share findings on append-only blackboard
  • Consensus Voting: Agents vote when analysis is complete
  • Local-First Processing: All processing happens client-side until explicit save

📚 Build Books

Curated organizational templates:

  • Standards Collections: Pre-packaged compliance standards
  • Tech Stack Bundles: Recommended technology combinations
  • Organization Branding: Custom covers and descriptions
  • Apply to Projects: One-click application of standards and tech stacks
  • Admin Publishing: Draft/Published workflow for admins
  • Gallery Discovery: Browse and clone published Build Books

🤝 Collaborative Document Editing

Real-time AI-assisted document editing:

  • CollaborationAgent: AI partner for document refinement
  • Line-Level Operations: Precise edit_lines with surgical changes
  • Project Context Attachment: Include requirements, canvas, artifacts as context
  • Change Tracking: Real-time diff visualization with timeline
  • Human-First Approach: AI respects human edits, never replaces entire documents
  • Merge to Artifact: Integrate changes back to original document

🖼️ Project Gallery

Discover and clone public projects:

  • Browse Published Projects: Explore community-shared projects
  • Preview Mode: View project details before cloning
  • One-Click Clone: Clone any public project to your dashboard
  • Search & Filter: Find projects by tags, name, or description

💻 AI Coding Agent

Autonomous file operations with full Git workflow:

  • Read, edit, create, delete, rename files
  • Staging → Commit → Push workflow
  • Real-time progress monitoring with operation timeline
  • Support for pause/resume and abort operations
  • Raw LLM logs viewer for debugging

⚡ Instant Collaboration

No-login-required sharing:

  • Token-based project access with role hierarchy
  • Anonymous project creation with session persistence
  • Real-time Supabase subscriptions for live updates
  • Token Management UI for owners

AI Agents & Orchestration

🤖 Multi-Agent AI Teams

Orchestrated AI specialists working together on canvas design:

Agent Focus
Architect Agent System structure and component hierarchy
Developer Agent Implementation details and code patterns
DBA Agent Database schemas and data modeling
Security Agent Vulnerability assessment and secure patterns
QA Agent Test coverage and validation strategies
DevOps Agent Deployment and infrastructure
UX Agent User experience and interface design
API Agent API design and integration patterns
Performance Agent Optimization and scalability
Documentation Agent Technical documentation

Agents share a blackboard for iterative refinement across multiple epochs with critic review.

📝 Specification Agents

Generate comprehensive documents from project data:

Agent Output
Overview Executive summary and project scope
Technical Specification Detailed technical documentation
Cloud Architecture AWS/Azure/GCP infrastructure design
API Specification OpenAPI/REST endpoint documentation
Security Analysis Threat modeling and security controls
Data Requirements Data models and storage requirements
Accessibility WCAG compliance and accessibility audit
Internationalization i18n/l10n requirements
DevOps CI/CD and deployment strategies
Testing Test strategy and coverage plans
Standards Compliance Regulatory compliance mapping
Executive Summary Business-focused project overview
Project Charter Governance and stakeholder documentation

🔍 Audit Agents

Multi-perspective analysis for cross-comparison:

Agent Perspective
Security Analyst Vulnerability and risk assessment
Business Analyst Business value and ROI analysis
Developer Implementation feasibility
End User Usability and user experience
Architect System design and patterns

Database Management

Pronghorn provides full PostgreSQL database lifecycle management with AI-powered data import.

🗄️ Provision & Connect

Feature Description
One-Click Provisioning Create PostgreSQL databases via Render.com with automatic configuration
External Connections Connect to any PostgreSQL instance (AWS RDS, Aurora, Neon, Railway, etc.)
SSL Configuration Support for require, prefer, disable SSL modes
Status Tracking Real-time database status (pending, creating, available, error, suspended)
Connection Testing Verify connectivity with 10-second timeout handling
Secure Storage Connection strings stored encrypted via edge function secrets

🔗 External Database Support

Connect to any PostgreSQL-compatible database:

  • AWS RDS/Aurora: Standard PostgreSQL connections with public accessibility
  • Neon/Supabase/Railway: Cloud PostgreSQL providers
  • Self-Hosted: Any PostgreSQL 12+ instance
  • Connection Timeout: 10-second timeout for unresponsive hosts
  • Network Guidance: Help for configuring security groups and VPC access

🔍 Schema Explorer

Browse and manage your database structure:

  • Tables - View columns, types, constraints, and indexes
  • Views - Materialized and standard views
  • Functions - PostgreSQL functions and procedures
  • Triggers - Database triggers with timing and events
  • Indexes - B-tree, hash, GIN, GiST indexes
  • Sequences - Auto-increment sequences
  • Types - Custom PostgreSQL types

📝 SQL Query Editor

Full-featured Monaco-powered SQL editor:

  • VS Code Engine - Syntax highlighting, auto-complete
  • Query Execution - Run queries with timing and result pagination
  • Saved Queries - Store frequently used queries per database
  • Query History - Access recent queries with keyboard shortcuts
  • Result Export - Export data as JSON, CSV, or SQL INSERT statements
  • Destructive Query Warnings - Visual indicators for DROP, DELETE, TRUNCATE

📥 Data Import Wizard

Multi-step wizard for importing data from files:

Step Description
1. Upload Drag-and-drop Excel (.xlsx, .xls), CSV, or JSON files
2. Preview View parsed data with automatic sheet/table detection
3. Schema AI-inferred or manual schema with type casting
4. Review SQL preview with batched INSERT statements
5. Execute Progress tracking with pause/resume capability

AI Schema Inference:

  • Automatic type detection (TEXT, INTEGER, BIGINT, NUMERIC, BOOLEAN, DATE, TIMESTAMP, JSONB)
  • Primary key recommendations
  • Index suggestions for common patterns
  • Foreign key relationship detection (JSON files)
  • Entity-Relationship Diagram (ERD) visualization

📋 Migration Tracking

Automatic DDL statement history:

  • CREATE - Tables, views, functions, indexes
  • ALTER - Column additions, type changes, constraints
  • DROP - Tracked for audit trail
  • Object Metadata - Schema, name, type for each migration
  • Execution Log - Timestamp, user, and full SQL content

Technology Stack

Frontend

Technology Purpose
React 18 UI framework with TypeScript
Vite Build tool and dev server
Tailwind CSS Utility-first styling
shadcn/ui Accessible component library
React Flow Interactive canvas diagrams
Monaco Editor Code editing (VS Code engine)
TanStack Query Server state management
React Router DOM Client-side routing
Framer Motion Animations and transitions
Recharts Data visualization and charts
jsPDF PDF generation
html-to-image Screenshot/image capture

Backend (Supabase)

Technology Purpose
PostgreSQL Primary database
Row Level Security Token-based access control
Edge Functions Deno serverless functions (53 functions)
Realtime WebSocket subscriptions
Storage File and artifact storage

LLM Providers

Provider Chat Models Image Models
Google Gemini gemini-2.5-flash, gemini-2.5-pro gemini-2.5-flash-image, gemini-3-pro-image-preview
Anthropic Claude claude-opus-4-5 -
xAI Grok grok-4-1-fast-reasoning, grok-4-1-fast-non-reasoning -

Project Structure

pronghorn/
├── src/
│   ├── components/
│   │   ├── ui/                    # shadcn/ui base components
│   │   ├── canvas/                # React Flow canvas components
│   │   │   ├── CanvasNode.tsx     # Node rendering
│   │   │   ├── CanvasPalette.tsx  # Node type selector
│   │   │   ├── AgentFlow.tsx      # Multi-agent orchestration UI
│   │   │   ├── AIArchitectDialog.tsx
│   │   │   ├── LayersManager.tsx
│   │   │   ├── Lasso.tsx
│   │   │   └── ...
│   │   ├── build/                 # Coding agent interface
│   │   │   ├── UnifiedAgentInterface.tsx
│   │   │   ├── AgentProgressMonitor.tsx
│   │   │   ├── StagingPanel.tsx
│   │   │   ├── RawLLMLogsViewer.tsx
│   │   │   └── ...
│   │   ├── deploy/                # Database & deployment components
│   │   │   ├── DatabaseExplorer.tsx      # Schema browser & SQL editor
│   │   │   ├── DatabaseImportWizard.tsx  # Multi-step data import
│   │   │   ├── SqlQueryEditor.tsx        # Monaco SQL editor
│   │   │   ├── ConnectDatabaseDialog.tsx # External DB connections
│   │   │   ├── ExternalDatabaseCard.tsx
│   │   │   ├── import/                   # Import wizard sub-components
│   │   │   │   ├── FileUploader.tsx
│   │   │   │   ├── ExcelDataGrid.tsx
│   │   │   │   ├── SchemaCreator.tsx
│   │   │   │   ├── DatabaseErdView.tsx
│   │   │   │   ├── JsonRelationshipFlow.tsx
│   │   │   │   └── SqlReviewPanel.tsx
│   │   │   └── ...
│   │   ├── present/               # Presentation generator components
│   │   │   ├── SlideRenderer.tsx
│   │   │   ├── SlideCanvas.tsx
│   │   │   ├── SlideThumbnails.tsx
│   │   │   ├── LayoutSelector.tsx
│   │   │   ├── FontScaleControl.tsx
│   │   │   ├── SlideImageGenerator.tsx
│   │   │   ├── PdfExportRenderer.tsx
│   │   │   └── ...
│   │   ├── audit/                 # Multi-agent audit components
│   │   │   ├── AuditConfigurationDialog.tsx
│   │   │   ├── AuditBlackboard.tsx
│   │   │   ├── TesseractVisualizer.tsx
│   │   │   ├── VennDiagramResults.tsx
│   │   │   ├── KnowledgeGraph.tsx
│   │   │   ├── KnowledgeGraphWebGL.tsx
│   │   │   ├── PipelineActivityStream.tsx
│   │   │   └── ...
│   │   ├── collaboration/         # Collaborative editing
│   │   │   ├── CollaborationEditor.tsx
│   │   │   ├── CollaborationChat.tsx
│   │   │   ├── CollaborationTimeline.tsx
│   │   │   ├── CollaborationHeatmap.tsx
│   │   │   └── ...
│   │   ├── buildbook/             # Build Book components
│   │   │   ├── BuildBookCard.tsx
│   │   │   ├── BuildBookChat.tsx
│   │   │   ├── ApplyBuildBookDialog.tsx
│   │   │   └── ...
│   │   ├── gallery/               # Project gallery components
│   │   │   ├── GalleryCard.tsx
│   │   │   ├── GalleryPreviewDialog.tsx
│   │   │   ├── GalleryCloneDialog.tsx
│   │   │   └── ...
│   │   ├── artifacts/             # Artifact management
│   │   │   ├── ArtifactPdfViewer.tsx
│   │   │   ├── ArtifactDocxViewer.tsx
│   │   │   ├── ArtifactExcelViewer.tsx
│   │   │   ├── VisualRecognitionDialog.tsx
│   │   │   ├── EnhanceImageDialog.tsx
│   │   │   └── ...
│   │   ├── repository/            # File tree, editor, Git integration
│   │   │   ├── FileTree.tsx
│   │   │   ├── CodeEditor.tsx
│   │   │   ├── IDEModal.tsx
│   │   │   ├── ContentSearchDialog.tsx
│   │   │   └── ...
│   │   ├── requirements/          # Requirements tree management
│   │   ├── standards/             # Standards library UI
│   │   ├── specifications/        # Specification generation
│   │   ├── dashboard/             # Project cards, creation dialogs
│   │   ├── layout/                # Navigation, sidebar, header
│   │   └── project/               # Project-specific selectors
│   │       ├── TokenManagement.tsx      # Token CRUD UI
│   │       ├── AccessLevelBanner.tsx
│   │       └── ...
│   │
│   ├── contexts/
│   │   ├── AuthContext.tsx        # Authentication state & methods
│   │   └── AdminContext.tsx       # Admin mode management
│   │
│   ├── hooks/
│   │   ├── useShareToken.ts       # Token extraction & caching
│   │   ├── useAuditPipeline.ts    # Audit orchestration state
│   │   ├── useRealtimeCanvas.ts   # Canvas real-time sync
│   │   ├── useRealtimeRequirements.ts
│   │   ├── useRealtimeArtifacts.ts
│   │   ├── useRealtimeLayers.ts
│   │   ├── useRealtimeCollaboration.ts
│   │   ├── useRealtimeBuildBooks.ts
│   │   ├── useInfiniteAgentMessages.ts
│   │   ├── useInfiniteAgentOperations.ts
│   │   └── ...
│   │
│   ├── pages/
│   │   ├── Landing.tsx            # Marketing landing page
│   │   ├── Dashboard.tsx          # Project list
│   │   ├── Auth.tsx               # Login/signup/SSO
│   │   ├── Gallery.tsx            # Public project gallery
│   │   ├── BuildBooks.tsx         # Build Book catalog
│   │   ├── BuildBookDetail.tsx    # Build Book viewer
│   │   ├── BuildBookEditor.tsx    # Build Book editor (admin)
│   │   ├── Standards.tsx          # Global standards library
│   │   ├── TechStacks.tsx         # Tech stack management
│   │   ├── Settings.tsx           # User settings
│   │   ├── Terms.tsx              # Terms of Use
│   │   ├── Privacy.tsx            # Privacy Policy
│   │   └── project/               # Project-specific pages
│   │       ├── Requirements.tsx
│   │       ├── Canvas.tsx
│   │       ├── Build.tsx
│   │       ├── Repository.tsx
│   │       ├── Artifacts.tsx
│   │       ├── Chat.tsx
│   │       ├── Deploy.tsx
│   │       ├── Database.tsx
│   │       ├── Audit.tsx
│   │       ├── Present.tsx
│   │       ├── Specifications.tsx
│   │       ├── Standards.tsx
│   │       └── ProjectSettings.tsx
│   │
│   ├── integrations/
│   │   └── supabase/
│   │       ├── client.ts          # Supabase client singleton
│   │       └── types.ts           # Generated TypeScript types
│   │
│   ├── lib/
│   │   ├── tokenCache.ts          # Synchronous token caching
│   │   ├── connectionLogic.ts     # Canvas edge validation
│   │   ├── stagingOperations.ts   # Git staging utilities
│   │   ├── presentationPdfExport.ts
│   │   ├── sqlParser.ts           # SQL parsing utilities
│   │   └── utils.ts               # Utility functions
│   │
│   └── main.tsx                   # Application entry point
│
├── supabase/
│   ├── functions/                 # 53 Deno edge functions (see below)
│   └── config.toml                # Supabase configuration
│
├── public/
│   ├── data/
│   │   ├── agents.json                    # Specification agents (13 types)
│   │   ├── buildAgents.json               # Canvas multi-agent definitions (10 agents)
│   │   ├── connectionLogic.json           # Canvas edge validation rules
│   │   ├── graphicStyles.json             # Image generation styles (6 categories)
│   │   ├── presentAgentInstructions.json  # Presentation agent blackboard spec
│   │   ├── presentationLayouts.json       # 15 slide layouts + themes
│   │   ├── auditAgentInstructions.json    # Audit orchestrator specification
│   │   ├── codingAgentInstructions.json   # Coding agent tools & patterns
│   │   ├── collaborationAgentInstructions.json  # Document collaboration agent
│   │   └── deploymentSettings.json        # Multi-runtime deploy configurations
│   │
│   └── features/
│       └── audit.md                       # Audit system documentation
│
└── README.md

Authentication System

Pronghorn implements a dual access model supporting both authenticated users and anonymous collaboration.

Authentication Methods

Method Description
Email/Password Traditional signup and login
Google SSO OAuth 2.0 redirect flow
Microsoft Azure SSO OAuth 2.0 with Azure AD
Anonymous Token-based access (no login required)

Auth Flow Architecture

┌─────────────────────────────────────────────────────────────────┐
│                      AuthContext Provider                        │
├─────────────────────────────────────────────────────────────────┤
│                                                                  │
│  ┌──────────────────┐    ┌──────────────────┐                   │
│  │  Email/Password  │    │   Social SSO     │                   │
│  ├──────────────────┤    ├──────────────────┤                   │
│  │ signUp()         │    │ signInWithGoogle │                   │
│  │ signIn()         │    │ signInWithAzure  │                   │
│  └────────┬─────────┘    └────────┬─────────┘                   │
│           │                       │                              │
│           └───────────┬───────────┘                              │
│                       ▼                                          │
│            ┌─────────────────────┐                               │
│            │  Supabase Auth      │                               │
│            │  onAuthStateChange  │                               │
│            └──────────┬──────────┘                               │
│                       ▼                                          │
│            ┌─────────────────────┐                               │
│            │  Session + User     │                               │
│            │  State Updated      │                               │
│            └─────────────────────┘                               │
│                                                                  │
└─────────────────────────────────────────────────────────────────┘

SSO Configuration

Google OAuth:

await supabase.auth.signInWithOAuth({
  provider: 'google',
  options: {
    redirectTo: `${window.location.origin}/dashboard`,
    skipBrowserRedirect: false  // Forces full page redirect
  }
});

Microsoft Azure:

await supabase.auth.signInWithOAuth({
  provider: 'azure',
  options: {
    scopes: 'openid profile email',
    redirectTo: `${window.location.origin}/dashboard`,
    skipBrowserRedirect: false
  }
});

The callback flow:

  1. User clicks SSO button → Redirects to provider
  2. Provider authenticates → Redirects to Supabase callback
  3. Supabase exchanges tokens → Redirects to /dashboard

Multi-Token RBAC System

Pronghorn implements a sophisticated role-based access control system using project tokens.

Token Architecture

-- project_tokens table
CREATE TABLE project_tokens (
  id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
  project_id UUID REFERENCES projects(id) ON DELETE CASCADE,
  token UUID NOT NULL DEFAULT gen_random_uuid(),
  role project_token_role NOT NULL DEFAULT 'viewer',
  label TEXT,                    -- Human-readable name
  created_at TIMESTAMPTZ DEFAULT now(),
  created_by UUID REFERENCES auth.users(id),
  expires_at TIMESTAMPTZ,        -- Optional expiration
  last_used_at TIMESTAMPTZ,      -- Usage tracking
  UNIQUE(token)
);

-- Role hierarchy
CREATE TYPE project_token_role AS ENUM ('owner', 'editor', 'viewer');

Role Permissions

Role Permissions
Owner Full access: manage tokens, delete project, all CRUD operations
Editor Create, read, update operations (no token management or deletion)
Viewer Read-only access to all project data

URL Pattern

/project/{projectId}/{page}/t/{token}

Examples:
/project/abc123/canvas/t/def456
/project/abc123/requirements/t/def456
/project/abc123/build/t/def456

Token Management UI

Manage project access tokens from Project Settings:

  • Create Tokens: Generate new viewer, editor, or owner tokens with optional labels
  • Set Expiration: Optional token expiry dates
  • Copy Share Links: One-click copy of shareable URLs
  • Roll Tokens: Regenerate token values while keeping the same ID
  • Revoke Access: Delete tokens to remove access
  • Current Token Indicator: Shows which token is currently in use

Core Authorization Functions

1. authorize_project_access - Validates access and returns role:

CREATE FUNCTION authorize_project_access(
  p_project_id UUID,
  p_token UUID DEFAULT NULL
) RETURNS project_token_role AS $$
BEGIN
  -- Check 1: Authenticated owner
  IF auth.uid() IS NOT NULL THEN
    IF EXISTS (SELECT 1 FROM projects WHERE id = p_project_id AND created_by = auth.uid()) THEN
      RETURN 'owner';
    END IF;
  END IF;
  
  -- Check 2: Valid token in project_tokens
  IF p_token IS NOT NULL THEN
    -- Return role from project_tokens if valid and not expired
    ...
  END IF;
  
  RAISE EXCEPTION 'Access denied';
END;
$$;

2. require_role - Enforces minimum permission level:

CREATE FUNCTION require_role(
  p_project_id UUID,
  p_token UUID,
  p_min_role project_token_role
) RETURNS project_token_role AS $$
DECLARE
  v_current_role project_token_role;
BEGIN
  v_current_role := authorize_project_access(p_project_id, p_token);
  
  -- Role hierarchy: owner(3) > editor(2) > viewer(1)
  IF role_to_level(v_current_role) < role_to_level(p_min_role) THEN
    RAISE EXCEPTION 'Insufficient permissions';
  END IF;
  
  RETURN v_current_role;
END;
$$;

RPC Patterns

All database operations use SECURITY DEFINER RPC functions with token validation.

Client-Side Pattern

import { supabase } from "@/integrations/supabase/client";
import { useShareToken } from "@/hooks/useShareToken";

function MyComponent({ projectId }: { projectId: string }) {
  const { token: shareToken, isTokenSet } = useShareToken(projectId);
  
  const loadData = async () => {
    // Wait for token to be ready
    if (!isTokenSet) return;
    
    const { data, error } = await supabase.rpc('get_requirements_with_token', {
      p_project_id: projectId,
      p_token: shareToken || null  // null for authenticated owners
    });
  };
}

RPC Function Structure

-- Read operation (requires viewer role)
CREATE FUNCTION get_requirements_with_token(
  p_project_id UUID,
  p_token UUID DEFAULT NULL
) RETURNS SETOF requirements
LANGUAGE plpgsql
SECURITY DEFINER
SET search_path TO 'public'
AS $$
BEGIN
  -- Validate access - require at least viewer role
  PERFORM require_role(p_project_id, p_token, 'viewer');
  
  RETURN QUERY 
    SELECT * FROM requirements 
    WHERE project_id = p_project_id
    ORDER BY order_index;
END;
$$;

-- Write operation (requires editor role)
CREATE FUNCTION insert_requirement_with_token(
  p_project_id UUID,
  p_token UUID,
  p_title TEXT,
  p_type requirement_type,
  p_parent_id UUID DEFAULT NULL
) RETURNS requirements
LANGUAGE plpgsql
SECURITY DEFINER
SET search_path TO 'public'
AS $$
DECLARE
  result requirements;
BEGIN
  -- Validate access - require at least editor role
  PERFORM require_role(p_project_id, p_token, 'editor');
  
  INSERT INTO requirements (project_id, title, type, parent_id)
  VALUES (p_project_id, p_title, p_type, p_parent_id)
  RETURNING * INTO result;
  
  RETURN result;
END;
$$;

Token Caching

Tokens are cached in memory for synchronous access:

// src/lib/tokenCache.ts
const tokenCache = new Map<string, string>();

export function setProjectToken(projectId: string, token: string): void {
  tokenCache.set(projectId, token);
}

export function getProjectToken(projectId: string): string | null {
  return tokenCache.get(projectId) || null;
}

export function clearProjectToken(projectId: string): void {
  tokenCache.delete(projectId);
}

Edge Functions

Pronghorn includes 53 Deno edge functions for server-side operations.

Project Management

Function Purpose
create-project Project creation with token generation
delete-project Secure project deletion with cascade
clone-project Full project cloning with all data
project-activity Activity analytics by time period
log-activity Activity event logging

AI Agents & Orchestration

Function Purpose
orchestrate-agents Multi-agent canvas design iteration
ai-architect Architecture generation
ai-architect-critic Architecture review and critique
coding-agent-orchestrator Autonomous coding agent with file operations
presentation-agent Blackboard-based slide generation
collaboration-agent-orchestrator Collaborative document editing
audit-orchestrator Multi-agent audit coordination
audit-extract-concepts Concept extraction for audit datasets
audit-merge-concepts Concept merging across datasets
audit-merge-concepts-v2 Enhanced concept merging with graph linking
audit-generate-venn Venn diagram synthesis from knowledge graph
audit-build-tesseract Evidence grid construction
audit-enhanced-sort Smart sorting for audit data

Requirements & Standards

Function Purpose
decompose-requirements AI decomposition into Epics/Features/Stories
expand-requirement Expand single requirement with AI
expand-standards Generate standards from descriptions
ai-create-standards Bulk AI standards generation
generate-specification Specification document generation

Chat & Streaming

Function Purpose
chat-stream-gemini Gemini streaming chat with SSE
chat-stream-anthropic Claude streaming chat with SSE
chat-stream-xai Grok streaming chat with SSE
summarize-chat Chat session summarization
summarize-artifact Artifact content summarization

Image & Media

Function Purpose
generate-image AI image generation (Gemini)
enhance-image Image editing/creation with Gemini image models
upload-artifact-image Image upload to Supabase storage
visual-recognition OCR and document text extraction
ingest-artifacts Bulk artifact ingestion from files

Repository & Git

Function Purpose
sync-repo-push Push commits to GitHub
sync-repo-pull Pull from GitHub
create-empty-repo Create empty repository
create-repo-from-template Clone from GitHub template
clone-public-repo Clone any public repository
link-existing-repo Link existing GitHub repo
staging-operations Stage/unstage/commit workflow

Database Management

Function Purpose
manage-database Schema browsing, SQL execution, data export
render-database Render.com PostgreSQL provisioning
database-agent-import AI-powered schema inference for imports
database-connection-secrets Secure connection string encryption

Deployment

Function Purpose
render-service Render.com web service management
deployment-secrets Deployment environment variable storage
generate-local-package Local development package generation
report-local-issue Local testing log capture and reporting

Presentation

Function Purpose
presentation-agent Blackboard-based slide generation
recast-slide-layout Slide layout restructuring with AI

Admin & Auth

Function Purpose
admin-management Admin role management
send-auth-email Custom auth emails via Resend
superadmin-github-management GitHub organization management
superadmin-render-management Render.com organization management

Edge Function Pattern

// supabase/functions/my-function/index.ts
import { serve } from "https://deno.land/std@0.168.0/http/server.ts";
import { createClient } from "https://esm.sh/@supabase/supabase-js@2";

const corsHeaders = {
  'Access-Control-Allow-Origin': '*',
  'Access-Control-Allow-Headers': 'authorization, x-client-info, apikey, content-type',
};

serve(async (req) => {
  // Handle CORS preflight
  if (req.method === 'OPTIONS') {
    return new Response(null, { headers: corsHeaders });
  }

  try {
    const { projectId, shareToken, ...params } = await req.json();
    
    // Create Supabase client with auth header
    const authHeader = req.headers.get('Authorization');
    const supabase = createClient(
      Deno.env.get('SUPABASE_URL')!,
      Deno.env.get('SUPABASE_ANON_KEY')!,
      { global: { headers: authHeader ? { Authorization: authHeader } : {} } }
    );
    
    // Validate access via RPC
    const { data: role, error: authError } = await supabase.rpc(
      'authorize_project_access',
      { p_project_id: projectId, p_token: shareToken || null }
    );
    
    if (authError || !role) {
      return new Response(JSON.stringify({ error: 'Access denied' }), {
        status: 403,
        headers: { ...corsHeaders, 'Content-Type': 'application/json' }
      });
    }
    
    // Perform operation...
    
    return new Response(JSON.stringify({ success: true }), {
      headers: { ...corsHeaders, 'Content-Type': 'application/json' }
    });
    
  } catch (error) {
    return new Response(JSON.stringify({ error: error.message }), {
      status: 500,
      headers: { ...corsHeaders, 'Content-Type': 'application/json' }
    });
  }
});

Real-Time Subscriptions

Pronghorn uses Supabase Realtime for live collaboration.

Subscription Pattern

import { useEffect, useRef } from 'react';
import { RealtimeChannel } from '@supabase/supabase-js';
import { supabase } from '@/integrations/supabase/client';

export function useRealtimeCanvas(projectId: string, shareToken: string | null) {
  const channelRef = useRef<RealtimeChannel | null>(null);
  const [nodes, setNodes] = useState([]);

  useEffect(() => {
    // Subscribe to changes
    channelRef.current = supabase
      .channel(`canvas:${projectId}`)
      .on(
        'postgres_changes',
        {
          event: '*',
          schema: 'public',
          table: 'canvas_nodes',
          filter: `project_id=eq.${projectId}`
        },
        (payload) => {
          // Handle INSERT, UPDATE, DELETE
          if (payload.eventType === 'INSERT') {
            setNodes(prev => [...prev, payload.new]);
          } else if (payload.eventType === 'UPDATE') {
            setNodes(prev => prev.map(n => 
              n.id === payload.new.id ? payload.new : n
            ));
          } else if (payload.eventType === 'DELETE') {
            setNodes(prev => prev.filter(n => n.id !== payload.old.id));
          }
        }
      )
      .on('broadcast', { event: 'canvas_refresh' }, () => {
        // Reload all data on broadcast
        loadCanvasData();
      })
      .subscribe((status) => {
        console.log('Subscription status:', status);
      });

    // Cleanup
    return () => {
      channelRef.current?.unsubscribe();
      channelRef.current = null;
    };
  }, [projectId]);

  // Broadcast changes to other clients
  const broadcastRefresh = () => {
    channelRef.current?.send({
      type: 'broadcast',
      event: 'canvas_refresh',
      payload: {}
    });
  };

  return { nodes, broadcastRefresh };
}

Key Patterns

  1. Use useRef for channel storage - Prevents stale closures
  2. Store channel during subscription - Required for broadcasting
  3. Use channelRef.current.send() - Not supabase.channel().send()
  4. Clean up on unmount - Unsubscribe and null the ref

Security Model

Pronghorn's real-time subscriptions use two distinct security models:

1. postgres_changes Events (RLS-Protected)

Database change events are fully secured by Row Level Security (RLS):

Security Layer Description
Token Validation set_share_token() RPC call configures session context
RLS Policies All tables have policies that check app.share_token
Server-Side Filtering Supabase only sends events the client is authorized to see 
// Before subscribing, the token is set in the session
await supabase.rpc('set_share_token', { token: shareToken });

// postgres_changes events respect RLS - unauthorized clients receive nothing
.on('postgres_changes', { 
  event: '*', 
  schema: 'public', 
  table: 'canvas_nodes',
  filter: `project_id=eq.${projectId}`  // Server enforces this + RLS
}, handleChange)

Security Guarantee: A client with only the projectId but no valid share_token cannot receive postgres_changes events.

2. Broadcast Events (Public Channels)

Broadcast channels are intentionally not private:

Aspect Status
Channel Privacy Public (no private: true flag)
Data Sensitivity None - broadcasts carry only refresh signals
Actual Data Access Still requires valid token via RPC calls 
// Broadcast sends NO sensitive data - just a refresh signal
channelRef.current?.send({
  type: 'broadcast',
  event: 'canvas_refresh',
  payload: {}  // Empty payload - no data exposed
});

// Client must still call RPC with valid token to get actual data
const { data } = await supabase.rpc('get_canvas_nodes_with_token', {
  p_project_id: projectId,
  p_token: shareToken  // Token validated server-side
});

Security Analysis:

  • Someone knowing only the projectId could technically subscribe to broadcast channels
  • However, they would only receive "refresh" signals with empty payloads
  • All actual data fetching requires a valid share_token validated by RLS
  • Risk Assessment: Minimal - no data leakage occurs

Summary: Defense in Depth

┌─────────────────────────────────────────────────────────────────┐
│                    Real-Time Security Layers                     │
├─────────────────────────────────────────────────────────────────┤
│                                                                  │
│  ┌─────────────────────────────────────────────────────────┐    │
│  │ postgres_changes                                         │    │
│  │ ✅ RLS-protected via share_token                        │    │
│  │ ✅ Server-side filtering                                │    │
│  │ ✅ No unauthorized data access possible                 │    │
│  └─────────────────────────────────────────────────────────┘    │
│                                                                  │
│  ┌─────────────────────────────────────────────────────────┐    │
│  │ Broadcast Events                                         │    │
│  │ ⚠️  Public channels (by design)                         │    │
│  │ ✅ Zero sensitive data in payloads                      │    │
│  │ ✅ Data fetching still requires valid token             │    │
│  └─────────────────────────────────────────────────────────┘    │
│                                                                  │
│  ┌─────────────────────────────────────────────────────────┐    │
│  │ Data Access (RPC Functions)                              │    │
│  │ ✅ All *_with_token functions validate share_token      │    │
│  │ ✅ SECURITY DEFINER with controlled search_path         │    │
│  │ ✅ Role-based permission checks                         │    │
│  └─────────────────────────────────────────────────────────┘    │
│                                                                  │
└─────────────────────────────────────────────────────────────────┘

Getting Started

Prerequisites

  • Node.js 18+
  • npm or bun

Installation

# Clone the repository
git clone https://github.com/pronghorn-red/pronghorn.git
cd pronghorn

# Install dependencies
npm install

# Start development server
npm run dev

Environment

The Supabase configuration is embedded in the client. No .env file is required for the frontend.

For edge functions, the following secrets are configured in Supabase:

Secret Purpose
GEMINI_API_KEY Google Gemini API access
ANTHROPIC_API_KEY Anthropic Claude API access
GROK_API_KEY xAI Grok API access
GITHUB_PAT Default repository operations
RENDER_API_KEY Render.com deployments & databases
RENDER_OWNER_ID Render.com account ID
RESEND_API_KEY Custom email delivery

Database Tables

Key tables for database management:

Table Purpose
project_databases Render.com hosted PostgreSQL instances
project_database_connections External database connections
project_database_sql Saved SQL queries per database
project_migrations DDL migration history tracking
project_tokens Project access tokens with roles

Deployment

Frontend

The frontend is hosted on Lovable at https://pronghorn.red.

To deploy updates:

  1. Push changes to the repository
  2. Lovable automatically builds and deploys

Backend (Edge Functions)

Edge functions deploy automatically when code is pushed. No manual deployment required.

Render.com (Optional)

For application deployments, Pronghorn supports Render.com:

Environment URL Pattern
Development dev-{appname}.onrender.com
Staging uat-{appname}.onrender.com
Production prod-{appname}.onrender.com

Supported Runtimes

Runtime Description
Node.js React, Vue, Express backends
Python Flask, FastAPI, Django
Go Compiled Go applications
Ruby Rails, Sinatra applications
Rust Compiled Rust binaries
Elixir Phoenix applications
Docker Custom Dockerfile deployments

Local Development Package

Generate a local development package for testing:

# Download package from Deploy page
# Extract and run:
npm install
npm start

The package includes:

  • pronghorn-runner.js - Watches files and auto-rebuilds
  • Telemetry integration with pronghorn.red
  • Environment configuration

Legal

Alpha Notice

This application is currently in Alpha testing by the Government of Alberta. Features, functionality, and availability are subject to change or may be removed at any time during the testing period.

Liability Waiver

By using this application, you acknowledge that it is provided "as is" without any warranties, express or implied. The Government of Alberta assumes no liability for any issues, data loss, or damages that may result from using this application during the testing period.

License

MIT License - See LICENSE for details.

Contact

Government of Alberta
Ministry of Technology and Innovation

About

No description, website, or topics provided.

Resources

Readme

License

MIT license
Activity
Custom properties

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

No packages published

Languages

  • TypeScript 87.5%
  • PLpgSQL 12.3%
  • Other 0.2%