Roadmap

Current version: 0.2.0

Foundation

Feature	Status	Details
Database initialization	Shipped	PostgreSQL via Drizzle ORM with auto-migration on startup
Authentication	Shipped	BetterAuth with email/password, session cookies, API tokens
Skills system	Shipped	Python skill scripts callable from Claude Code hooks

Core Features

Feature	Status	Details
Task management	Shipped	Full CRUD with projects, priorities, tags, soft delete, and restore
Project management	Shipped	Full CRUD, soft delete, color coding, task/document association
Cron agents	Shipped	Scheduled Claude Agent SDK execution with budget caps and run history
Interactive chat	Shipped	WebSocket streaming chat with Claude, conversation persistence, cost tracking
CLI installer	Shipped	`cognova init`, `update`, `start`, `stop`, `restart`, `reset`, `doctor`
Memory system	Shipped	Extraction from conversations, session injection, access tracking, search + skill
Document metadata	Shipped	Vault file sync with tags, sharing, project association, frontmatter parsing
Document sharing	Shipped	Public/private sharing with UUID-based access and SEO control
Web terminal	Shipped	Full PTY terminal via xterm.js and node-pty with session persistence
File system API	Shipped	Vault CRUD with path security validation
Dashboard	Shipped	Stat cards, upcoming tasks, recent chats/docs, usage summary
Secrets API	Shipped	AES-256-GCM encrypted key-value store with settings UI
Settings page	Shipped	Profile, password, secrets management, notification preferences
Notifications	Shipped	Real-time WebSocket event bus across 9 resource types with per-action preferences
Hook analytics	Shipped	Claude Code hook event logging, stats, and filtering dashboard
Token usage tracking	Shipped	Per-source cost breakdown with daily/hourly granularity
Docker deployment	Shipped	Dockerfile, docker-compose, health checks, volume support

Planned

Feature	Status	Details
Full-text search	Planned	PostgreSQL `tsvector` indexing across vault files with relevance ranking
Secrets skill	Planned	Claude skill for agents to securely access secrets at runtime
Memory reinforcement	Planned	Relevance decay, access-based scoring, expiration cleanup
AI history export	Planned	JSONL transcript parsing for conversation export
Editor UX improvements	Planned	Slash commands, bubble toolbar, drag handles, syntax highlighting in code blocks
Keyboard shortcuts	Planned	Cmd+K search, shortcuts for common actions via Nuxt UI composables
Mobile responsive	Planned	`sm:`/`md:` breakpoint coverage (currently `lg:` only)

Future

Feature	Status	Details
Obsidian integration	Exploring	Plugin to upload and share documents directly into the vault
Semantic search	Exploring	pgvector embeddings for meaning-based search across vault and memory
Table support in editor	Exploring	TipTap table extensions for markdown editing
Image uploads	Exploring	Drag-and-drop image handling in the editor

Status Key

Icon	Meaning
Shipped	Implemented and working
Planned / Exploring	On the radar, not yet implemented

Dependency Graph

database-init ──┬── auth
                ├── task management
                ├── project management
                ├── search
                └── secrets API

skills system ──┬── task skill
                └── cron agents

secrets API ────── secrets skill (planned)

Database Migrations

How to modify the schema, generate migrations, and apply them in development and production.

Cognova - Open Source AI Agent Platform

A self-hosted AI knowledge workspace with scheduled agents, document management, encrypted secrets, and extensible skills.