PromptVault v2.2.0 - Official Apache 2.0 Open-Source License, API Diagnostics & Agentic Discovery

PromptVault 2.2 brings official Apache 2.0 open-source licensing, root NOTICE anti-theft protection, zero-token API key diagnostics, agentic web discoverability, performance optimizations, and refined UI controls.

Highlights

⚖️ Official Apache 2.0 License & NOTICE Protection (Anti-Theft Shield)

• Permissive Apache 2.0 License: Formally standardized under Apache 2.0, welcoming community forks, star velocity, and enterprise adoption with zero legal friction.
• Root NOTICE File (Section 4d): Legally mandates that any fork, redistribution, or hosted deployment preserve the root NOTICE file and display visible attribution to Sagar Biswas (@SagarBiswas-MultiHAT) within documentation and user interfaces.
• Trademark Protection (Section 6): Protects "PromptVault" and "MultiHAT" branding from unauthorized clones.
• SPDX Fingerprints Embedded: Added canonical machine-readable provenance headers across all application source files to prevent commit-wiped source theft.
• Persistent UI Attribution: Integrated author credits into the status bar footer and Settings modal linking to https://multihat.dev and canonical source.

🔍 Zero-Token API Key Health Diagnostic CLI

• Added npm run test:keys to instantly inspect your Gemini API key pool.
• Tests individual keys with zero token consumption against Google Generative Language endpoints.
• Provides real-time cooldown detection, rate-limit status, and an interactive --wait watch mode.

⚡ Server Resilience & Deduplication

• Added in-flight request deduplication to prevent redundant concurrent LLM calls.
• Integrated circuit-breaker triggers for authentication errors and expired keys.

🤖 Agentic Web Discovery & SEO

• Added llms.txt and llms-full.txt for AI agent crawlers (ChatGPT, Claude, Perplexity).
• Automated 1200x630 Open Graph social card generator (npm run generate:og).
• Added sitemap, robots.txt, Web App Manifest, and JSON-LD schema metadata connecting creator Sagar Biswas.

🚀 Performance & Critical Path Tuning

• Optimized Vite build output by excluding lazy Supabase and Motion chunks from the critical modulepreload list.
• Deferred non-critical script loading (GA4 analytics) and added responsive logo asset delivery (logo-sm.png).

🎨 UI & UX Refinements

• Custom dark sort dropdown with full keyboard navigation.
• Dynamic real-time online/offline connectivity indicator.
• Refactored settings modal with dedicated protocol shortcuts and backup controls.
• Prevented tag pill text wrapping and overlapping on prompt cards.

---

Upgrading from v2.1.0

No breaking changes or schema migrations. Run npm install and restart your dev or production server.

PromptVault v2.1.0 - Gemini multi-key pool and rate-limit circuit breakers

PromptVault 2.1 improves reliability and resilience of the AI Librarian when working with the Gemini API, and cleans up a few rough edges carried over from 2.0.

What is new

Gemini multi-key pool with per-key rate-limit circuit breakers

You can now supply more than one Gemini API key by adding numbered environment variables:

GEMINI_API_KEY_1=...
GEMINI_API_KEY_2=...
GEMINI_API_KEY_3=...

The proxy tries each key in order. When a key hits the Gemini 429 rate limit, the server reads the retry in Xs hint from the response body and skips that key for exactly that duration, then tries the next key immediately. Only when every Gemini key is exhausted does the chain fall through to Groq.

This means a single rate-limited key no longer blocks all requests for a full minute, and you can spread load across multiple free-tier keys without any manual rotation logic.

Circuit breakers also activate on timeouts and upstream 5xx errors, backing off the affected key for 60 seconds to avoid hammering an unhealthy endpoint.

The /api/health endpoint now reports a separate status entry for each key slot (e.g. gemini[0], gemini[1]) so you can see which keys are healthy and which are in cooldown.

Separate Groq token limit and input truncation

Groq and Gemini now have independent MAX_OUTPUT_TOKENS values in config.ts. Oversized prompts are truncated before being sent to Groq, preventing unnecessary rejections.

Fixes

• Resolved three TypeScript strict-mode errors in suggest.ts that caused CI to fail.

Refactoring

• Removed the custom CSS preload Vite plugin and its associated index.html placeholder. CSS is now handled by standard Vite behaviour.
• Aligned the esbuild compile target to es2022 in vite.config.ts to match the build target.

Documentation

• Added an Architecture Overview diagram to the README.

Upgrading from 2.0.0

No breaking changes. Existing single-key setups continue to work without modification. To use the new key pool, rename your existing GEMINI_API_KEY to GEMINI_API_KEY_1 and add further keys as needed. The old GEMINI_API_KEY name is still supported as a fallback for single-key deployments.

PromptVault v2.0.0 - Encrypted vaults, safer sync

PromptVault 2.0 is a major reliability and privacy update for people who want to keep their best AI prompts organised without losing control of their data.

The app is still simple at its core: save useful prompts, find them quickly, and reuse them whenever you need them. This release makes that experience safer, smoother, and much more resilient behind the scenes.

Highlights

• Real local vault encryption — protect prompts with AES-256-GCM encryption using either a PIN or a passphrase. Your secret is never stored, and a one-time recovery key is provided when encryption is enabled.
• Safer cloud sync — prompt and category changes are merged more carefully across devices, including deletions, instead of relying on a simple last-write-wins update.
• A more reliable AI Librarian — improved provider handling, request timeouts, structured responses, caching, and clearer errors make prompt analysis and rewriting more dependable.
• Better day-to-day UX — refreshed navigation, prompt views, settings, sorting, keyboard shortcuts, dialogs, and responsive layouts.
• Stronger production defaults — tighter request validation, rate limiting, CORS checks, security headers, and clearer deployment configuration.
• Quality checks built in — strict type checking, unit tests, end-to-end tests, and GitHub Actions CI now cover the app and server.

Upgrading

Existing local vaults remain available. After updating, choose Encrypt now from the app to protect an older vault with the new encryption system. Save the recovery key somewhere safe — it is shown only once.

For self-hosted deployments, compare your existing .env file with .env.example. In production, ALLOWED_ORIGINS is now required, and deployments behind a reverse proxy may need TRUST_PROXY_HOPS configured.

Thank you for trying PromptVault and for helping shape its next chapter.

PromptVault v1.0.0 - First Release

PromptVault is now ready for its first public release.

This release introduces a private, self-hosted library for saving and reusing your best AI prompts. It is built for people who want to keep useful prompts organised without giving up control of their data.

What’s included

• Save, edit, search, rate, and favourite prompts
• Custom categories and tags for easy organisation
• Reusable prompt templates with {{placeholders}}
• AI Librarian tools to score, improve, and organise prompts
• PIN protection for a more private local vault
• Backup and restore for your full prompt collection
• Optional Google sign-in and Supabase cloud sync
• Usage statistics, light/dark mode, and mobile-friendly layouts
• Gemini support with automatic Groq fallback

Getting started

Clone the repository, install dependencies, add your API keys to .env, then run the frontend and API server locally. Full setup and deployment instructions are available in the README.

Thanks for checking out PromptVault. I hope it makes keeping great prompts a little easier.