Lorapok Labs is an open-source software collective focused on Sensory Computing, Biological UI, and Autonomous AI Engineering. This portal serves as the primary gateway for 30+ open-source tools, browser extensions, developer CLIs, and autonomous research publications across 7 global distribution platforms.
Lorapok Labs operates on a multi-tier distributed architecture spanning an edge ingress network, client-side sensory SPA, static pre-rendering build pipeline, and an autonomous AI systems research engine (LoLaBo).
flowchart TD
subgraph Ingress ["π Ingress & Edge Layer"]
Users["π₯οΈ Web Visitors & Engineers"]
Bots["π€ Search Crawlers & Social Indexers"]
DiscordChannel["π¬ Discord Community (#announcements)"]
CF["π‘οΈ Cloudflare Global Edge Network<br/>β’ Custom Domain: lorapok.tech<br/>β’ SSL/TLS Termination & DDoS Shield<br/>β’ Edge Worker Proxies"]
GH_CDN["π¦ GitHub Pages Global CDN<br/>β’ Static Asset Edge Distribution<br/>β’ Geo-Distributed Cache"]
end
subgraph ClientApp ["π¨ Client Presentation Tier (React 19 SPA)"]
Router["π§ React Router v7<br/>β’ Route-Level Code Splitting (React.lazy)<br/>β’ URL Deep-Linking & History"]
subgraph Views ["Application Views"]
HomeView["π / (Home & Hero)"]
ProjectsView["π /projects (30+ Catalog & βK)"]
TeamView["π₯ /team (Interactive Dossiers)"]
AboutView["π /about (Mission & Philosophy)"]
DevPortal["π¬ /dev (DevModePortal & AI Lab)"]
BlogView["π° /blog & /blog/:slug (LoLaBo Portal)"]
end
ImgCache["β‘ Progressive Image Engine<br/>β’ In-Memory Session Cache (Set)<br/>β’ Low-Poly Blur-Up Skeletons<br/>β’ Runtime Dedup Guard"]
end
subgraph BuildPipeline ["βοΈ Pre-Rendering & Postbuild Engine"]
ViteCompiler["β‘ Vite 8 + Rollup<br/>β’ TypeScript Compilation<br/>β’ Tailwind v4 Token Shake"]
PostBuildScript["π οΈ scripts/postbuild.js<br/>β’ Pre-renders 7 Core HTML Routes<br/>β’ Pre-renders 19+ Static Blog SEO Pages<br/>β’ Injects Schema.org JSON-LD<br/>β’ Generates XML Sitemaps & RSS 2.0"]
end
subgraph LoLaBoAI ["π§ LoLaBo Autonomous AI Subsystem (agent/)"]
Trigger["β±οΈ Staggered Cron / CI Trigger<br/>(GitHub Actions: :17, :47)"]
KeyPool["π Multi-Account Key Rotator (keyManager.ts)<br/>β’ 4+ Google Account Pool<br/>β’ Smart Round-Robin & Instant 429 Failover"]
Validator["π‘οΈ Model Validator & Deprecation Guard (modelValidator.ts)<br/>β’ Real-Time Google API Discovery<br/>β’ Auto-Quarantine Deprecated Models (404/Retired)<br/>β’ Zero-Downtime Cascade"]
ModelTiering["βοΈ Multi-Tier Model Matrix (writer.ts)<br/>β’ Gemini 3.x Primary (3.8/3.7/3.6/3.1/3.5)<br/>β’ Gemini 2.x Resilient Fallbacks (2.5/2.0/Thinking)"]
Author["βοΈ Research Authoring Engine (writer.ts)<br/>β’ Zero-Truncation Completeness Barrier<br/>β’ ASCII Topologies, Benchmarks & Formal Proofs"]
PeerReview["π¬ Autonomous Review Unit (reviewer.ts)<br/>β’ 100-Point Academic Rubric<br/>β’ Dynamic Revision Feedback Loop (<85)"]
VisualEngine["π¨ Swiss Editorial Visual Suite (imageAgent.ts)<br/>β’ Hero Cover: Swiss / Bauhaus Minimalist Illustration<br/>β’ Fig 1: Native Code-to-Diagram Mermaid.js Flowchart<br/>β’ Fig 2: Crisp 100% Vector SVG Latency CDF Plot"]
Consolidator["πΎ Export & Normalizer (exportData.ts)<br/>β’ Zero-Duplicate Catalog Enforcement<br/>β’ Dual Persistence Normalizer"]
Verifier["π Production Edge Verifier (broadcastLivePost.ts)<br/>β’ Validates 200 OK on Edge<br/>β’ Atomic Discord Embed Dispatch"]
end
subgraph DataStorage ["ποΈ Multi-Tier Persistence"]
FirestoreDB[("π₯ Google Cloud Firestore<br/>β’ Canonical Cloud Store<br/>β’ blog_posts collection")]
StaticJSON[("π Static Catalog Baseline<br/>β’ public/blog/posts.json")]
ModelQuarantine[("π‘οΈ Deprecated Models Quarantine<br/>β’ .loragent-debug/deprecated-models.json")]
end
Users --> CF
Bots --> CF
CF --> GH_CDN
GH_CDN --> Router
Router --> HomeView
Router --> ProjectsView
Router --> TeamView
Router --> AboutView
Router --> DevPortal
Router --> BlogView
BlogView --> ImgCache
Trigger --> KeyPool
KeyPool --> Validator
Validator --> ModelTiering
ModelTiering --> Author
Author --> PeerReview
PeerReview -->|"Score < 85 (Revisions Required)"| Author
PeerReview -->|"Score >= 85 (Approved)"| VisualEngine
VisualEngine --> Consolidator
Validator -.-> ModelQuarantine
Consolidator --> FirestoreDB
Consolidator --> StaticJSON
StaticJSON --> PostBuildScript
ViteCompiler --> PostBuildScript
PostBuildScript --> GH_CDN
Consolidator --> Verifier
Verifier -->|"Live 200 OK Verified"| DiscordChannel
The sequence diagram below illustrates the end-to-end autonomous research synthesis, peer-review refinement, online AI visual generation, and gated edge broadcast:
sequenceDiagram
autonumber
actor Cron as β±οΈ Cron Runner (:17, :47)
participant KeyMgr as π KeyManager (4-Key Pool)
participant Validator as π‘οΈ Model Validator (modelValidator.ts)
participant Writer as βοΈ LoLaBo Authoring Engine
participant Reviewer as π¬ Autonomous Peer Review Unit
participant Visuals as π¨ Swiss Editorial Visual Suite
participant Export as πΎ Consolidator & Normalizer
participant Storage as ποΈ Firestore & posts.json
participant Deploy as π GitHub Actions Deploy
participant Probe as π Production Edge Probe
participant Discord as π¬ Discord Webhook
Cron->>KeyMgr: Request active Google AI Key (Pool 1..4)
KeyMgr-->>Writer: Yield healthy key (cooldown on 429 errors)
Writer->>Validator: Validate candidate models ladder (query /v1beta/models)
Validator-->>Writer: Return active candidate models (filter quarantined 404/deprecated)
alt Model returns 404 / NOT_FOUND / Deprecated
Writer->>Validator: Auto-prune model from candidate ladder & persist to quarantine
Writer->>Writer: Cascade immediately to next candidate model in ladder
end
alt Topic is Deep Systems Research, RFC, or Thesis
Writer->>Writer: Flagship Tier: Gemini 3.8/3.7/3.6 Flash β 3.1 Pro β Pro Latest β 2.5 Pro β 2.0 Thinking
else Topic is Standard Technical Dispatch or News
Writer->>Writer: Technical Tier: Gemini 3.8/3.7/3.6 Flash β Flash Latest β 2.5 Flash β 2.0 Flash
end
Writer->>Writer: Synthesize manuscript (Zero-Truncation Barrier, ASCII diagrams, formal math, benchmark tables)
loop Peer-Review Verification Loop (Max 2 Passes)
Writer->>Reviewer: Submit draft for evaluation
Reviewer->>Validator: Validate reviewer candidate models
Reviewer->>Reviewer: Audit against 100-Point Rubric (Rigor, Math, Code, Citations)
alt Score below 85 (Defects or Missing Proofs)
Reviewer-->>Writer: Reject with structured critique instructions
Writer->>Writer: Revise manuscript addressing critique
else Score 85 or above (Approved)
Reviewer-->>Writer: Issue peer-review approval badge
end
end
Writer->>Visuals: Request publication-grade visual suite (Swiss Cover, Fig 1 Mermaid, Fig 2 SVG)
Visuals->>Visuals: Synthesize Swiss flat vector cover, Mermaid architecture flowchart & vector latency CDF plot
Visuals-->>Writer: Yield complete visual suite & inject into markdown sections
Writer->>Export: Send verified post package
Export->>Storage: Atomically commit to Cloud Firestore & public/blog/posts.json
Storage->>Deploy: Trigger static build & pre-rendering (scripts/postbuild.js)
Deploy->>Deploy: Publish static pages & OpenGraph tags to Edge CDN
Deploy->>Probe: Poll https://lorapok.tech/blog/:slug until HTTP 200 OK
Probe-->>Deploy: Edge verification confirmed
Deploy->>Discord: Broadcast rich publication embed with verified live link
To permanently guarantee that no two articles ever share the same visual and eliminate stock photography fallbacks, LoLaBo enforces a three-layer defense matrix:
flowchart LR
subgraph Layer1 ["Layer 1: Generation Time (imageGen.ts)"]
A1["Authoring Engine"] --> A2["Extract Technical imagePrompt"]
A2 --> A3["Query usedImages Set"]
A3 --> A4["Calculate Prime Seed: (baseSeed + i * 104729)"]
A4 --> A5["Unique Pollinations AI Flux URL"]
end
subgraph Layer2 ["Layer 2: Export & Normalization (exportData.ts)"]
B1["Consolidator Process"] --> B2["Scan Merged Catalog"]
B2 --> B3{"Unsplash or Duplicate?"}
B3 -->|Yes| B4["Re-synthesize Online AI Flux Visual"]
B3 -->|No| B5["Retain Valid Visual"]
B4 --> B6["Sync Back to Cloud Firestore"]
B5 --> B7["Write to public/blog/posts.json"]
B6 --> B7
end
subgraph Layer3 ["Layer 3: Browser Runtime (BlogApp.tsx)"]
C1["Client Fetches posts.json / Live Sync"] --> C2["ensureUniqueVisuals() Guard"]
C2 --> C3{"In-Memory URL Conflict?"}
C3 -->|Yes| C4["On-the-Fly Prime Salt Resolution"]
C3 -->|No| C5["Render Progressive Image Component"]
C4 --> C5
end
Layer1 --> Layer2
Layer2 --> Layer3
Lorapok Labs v2.0 transitions the ecosystem from a single-page portfolio into a full-fledged, multi-page open-source product organization with route-level code splitting:
| Route | View Component | Key Architectural Capabilities |
|---|---|---|
/ |
HomePage.tsx |
Sensory hero, animated stats counter, 3-pillar philosophy, featured product spotlights. |
/projects |
ProjectsPage.tsx |
30+ products across 10 categories, multi-criteria filtering, keyboard βK search overlay. |
/team |
TeamPage.tsx |
Interactive dossiers, skills matrices, career trajectories, and modal deep dives. |
/about |
AboutPage.tsx |
Organizational backstory, founder profile, design system manifesto, and ethics. |
/changelog |
ChangelogPage.tsx |
Chronological release milestones, semantic versions, and platform launch notes. |
/support |
SupportPage.tsx |
bKash funding channel, multi-chain crypto addresses, and backer recognition. |
/contact |
ContactPage.tsx |
Direct routing contact bridge for commercial inquiries, partnerships, and bug reports. |
/dev |
DevModePortal.tsx |
Authenticated developer workspace, multi-provider AI lab, API playground, live system telemetry. |
/blog |
BlogApp.tsx |
LoLaBo research portal, category/format filters, progressive image caching, reading progress. |
/blog/:slug |
BlogApp.tsx |
Pre-rendered individual research treatise view, formal citations, LaTeX math, code highlighting. |
Connects to multiple Google AI Studio accounts to multiply rate limits and ensure zero publication downtime:
-
Environment Key Matrix: Accepts
GEMINI_API_KEYS="key1,key2,key3,key4"or individual keysGEMINI_API_KEY_1..4. -
Smart Failover: Catches HTTP
429 Too Many RequestsorRESOURCE_EXHAUSTEDresponses, enters a cooldown state for the exhausted account, and automatically fails over to the next healthy account in$< 5\text{ ms}$ .
Enforces a multi-tiered, latency- and depth-optimized fallback ladder across model families:
| Content Tier | Primary Gemini 3.x Models | Preserved Resilient 2.x Fallbacks | Target Depth & Invariants |
|---|---|---|---|
| Flagship Research Treatises & RFCs |
gemini-3.8-flash, gemini-3.7-flash, gemini-3.6-flash, gemini-flash-latest, gemini-3.1-pro-preview, gemini-pro-latest, gemini-3.5-flash, gemini-3.1-flash-lite, gemini-flash-lite-latest
|
gemini-2.5-pro, gemini-2.5-flash, gemini-2.0-flash, gemini-2.0-flash-lite, gemini-2.0-pro-exp-02-05, gemini-2.0-flash-thinking-exp-01-21, gemini-1.5-pro, gemini-1.5-flash
|
2,500β3,500 words, formal math ($O(n \log n)$), ASCII topologies, memory invariants, 4β6 peer-reviewed citations. |
| Technical Dispatches & Engineering Blogs |
gemini-3.8-flash, gemini-3.7-flash, gemini-3.6-flash, gemini-flash-latest, gemini-3.5-flash, gemini-3.1-flash-lite, gemini-flash-lite-latest
|
gemini-2.5-flash, gemini-2.0-flash, gemini-2.0-flash-lite, gemini-1.5-flash
|
1,400β2,200 words, working code implementations, low-level data-plane dynamics, actionable takeaways. |
To prevent stalled execution caused by retired, deprecated, or unavailable model aliases:
-
Real-Time Deprecation Interception: Monitors Google API responses for HTTP
404 NOT_FOUND,INVALID_ARGUMENT, and explicit deprecation signatures ("is not found for API version","is deprecated","no longer available","unsupported model"). -
Automated Quarantine & Pruning: Automatically purges deprecated models from in-memory candidate arrays and persists them to
.loragent-debug/deprecated-models.jsonso retries never waste quota on defunct endpoints. -
Remote Model Discovery: Automatically queries
GET https://generativelanguage.googleapis.com/v1beta/models?key=${apiKey}to verify activegenerateContentcapabilities against Google's live catalog. -
Zero-Downtime Cascade: If a model alias is rejected, the engine cascades to the next candidate model in
$< 10\text{ ms}$ without dropping the publication task.
Prevents half-done, truncated, or unclosed articles from reaching production:
- Integrity Validation: Rigorously inspects markdown code fences, unclosed backticks, truncated tables, incomplete formulas, and ensures mandatory
## Key Takeaways & Summaryand## References & Technical Citationsheaders. - Surgical Continuation Pass (
completeArticleSections): If token ceilings are encountered, a targeted completion agent cleanly closes severed code blocks and generates formal takeaways and academic references.
Replaces generic sci-fi tropes (octane render, glowing neon, cyberpunk, cybernetic, holographic, futuristic, laser lines) with publication-grade engineering graphics:
- Hero Cover (1200x630): Swiss / Bauhaus minimalist 2D vector technical illustration with clean graphite backdrops, muted slate/cobalt palettes, and precision line art.
-
Figure 1 (Architecture Blueprint): Authentic code-to-diagram Mermaid.js flowcharts (
flowchart TD) rendered natively with interactive expand, copy source, and dark themes. -
Figure 2 (Empirical Benchmarks): Publication-grade 100% Vector SVG latency CDF curves (
$P_{50}, P_{90}, P_{99}, P_{99.9}$ ) with microsecond coordinates and throughput distributions.
Audits all generated manuscripts against an academic 100-point rubric before publication:
| Evaluation Dimension | Weight | Acceptance Criteria |
|---|---|---|
| Architectural Depth & Rigor | 30 pts | Kernel-level mechanics, memory alignments, cache hierarchy impacts, and invariants. |
| Structural Completeness | 25 pts | ASCII topology diagrams, comparative benchmark matrices, and formal conclusion. |
| Code & Mathematical Soundness | 20 pts | Idiomatic Rust/Go/C++ snippets, asymptotic complexity bounds ($O(n \log n)$). |
| Citations & Literature Integrity | 15 pts | Inline citations referencing authoritative RFCs/papers, dedicated references section ( |
| Anti-Hallucination & Tone | 10 pts | Zero template placeholders (TODO, TBD), zero abrupt cutoffs, authoritative engineering voice. |
Drafts scoring $< 85$ receive targeted critique prompts and enter an iterative revision cycle (up to 2 passes).
| Content Tier | 1 Google Account | 4 Google Accounts Pooled | With Free Provider Mesh (Groq / Cerebras) |
|---|---|---|---|
| Standard Blogs & Dispatches | 1,500 / day (62.5/hr) | 6,000 / day (250/hr) | 20,000+ / day |
| Flagship Research & Thesis Papers | 50 / day (2.08/hr) | 200 / day (8.33/hr) | 500+ / day |
| Peer-Review Verification Passes | 1,000 / day | 4,000 / day | 10,000+ / day |
All visual assets in the LoLaBo blog portal utilize LoLaBoImage.tsx:
- In-Memory Session Cache: Tracks loaded images via
loadedImageCache: Set<string>, eliminating network re-fetches during internal client navigation. - Low-Poly Blur-Up Skeletons: Displays an animated SVG shimmer skeleton while assets decode.
- Hero Image Prefetching: Automatically prefetches hero images for the top articles upon initial portal hydration.
- Deterministic AI Fallback: If an image fails to load over the network, a dynamic vector SVG with technical glyphs and domain branding renders instantly.
To guarantee first-class indexability on search engines and instant social sharing previews without server-side rendering (SSR) overhead:
- Isolated Static HTML Routes: Pre-renders separate
index.htmlfiles for/projects,/about,/team,/changelog,/support,/contact, and/dev. - Dedicated Article SEO Pages: Generates static HTML folders for all 19+ research articles under
dist/blog/:slug/index.html. - Automated Metadata Injection: Injects page-specific
<title>,<meta name="description">,og:image,twitter:card, and canonical URLs. - Structured Data (JSON-LD): Injects Schema.org
ArticleandOrganizationstructured data into every page's<head>. - Sitemaps & RSS Feed: Automatically synthesizes
public/sitemap.xml,public/blog/sitemap.xml, andpublic/blog/rss.xml.
lorapok.github.io/
βββ agent/ # LoLaBo Autonomous AI Subsystem
β βββ index.ts # CLI entry point & publication orchestration
β βββ server.ts # Autonomous daemon & REST API service
β βββ writer.ts # Multi-model authoring engine & zero-truncation barrier
β βββ modelValidator.ts # Real-time model health & automated deprecation pruner
β βββ reviewer.ts # Autonomous Research Review Unit (100-pt rubric)
β βββ keyManager.ts # 4-account Google AI key rotator & 429 failover
β βββ imageAgent.ts # Swiss editorial visual synthesizer (Mermaid & vector SVG)
β βββ imageGen.ts # 100% online AI Flux engine & prompt normalizer
β βββ exportData.ts # Static data consolidator & Firestore synchronizer
β βββ broadcastLivePost.ts # Edge verification probe & Discord broadcaster
β βββ capacity.ts # Multi-account publishing capacity analyzer
βββ public/ # Static Web Assets
β βββ blog/
β β βββ posts.json # Canonical static research catalog (19 articles)
β β βββ sitemap.xml # Blog XML sitemap
β β βββ rss.xml # RSS 2.0 publication feed
β βββ sitemap.xml # Root website XML sitemap
β βββ robots.txt # Search engine crawler policies
βββ scripts/ # Build & Release Tooling
β βββ postbuild.js # Static HTML pre-renderer & SEO tag injector
βββ src/ # React 19 Client SPA Source
β βββ blog/ # LoLaBo Research Portal
β β βββ BlogApp.tsx # Main blog application & Mermaid.js renderer
β β βββ components/ # Blog components (LoLaBoImage, CodeBlocks, etc.)
β βββ components/ # Shared UI Components (Navbar, Footer, Search)
β βββ data/ # Structured Ecosystem Data
β β βββ lorapok.ts # 30+ products, categories, and brand metadata
β β βββ team.ts # Core team profiles & interactive dossiers
β β βββ changelog.ts # Ecosystem version milestones
β βββ dev/ # Developer Workspace & AI Playground
β βββ pages/ # Core Route Page Views
β βββ styles/ # Tailwind CSS v4 & custom design tokens
βββ .github/workflows/ # Continuous Integration & Automation
β βββ lolabo-agent.yml # Twice-hourly autonomous research publisher
β βββ deploy.yml # Automated edge build & GitHub Pages deploy
βββ package.json # Project dependencies & operational scripts
Lorapok products are distributed across 7 major software registries:
- VS Code Marketplace: LorapokLabs β Cursor Curse Monitor, Lorapok Atlas
- Open VSX Registry: LorapokLabs β Cursor Curse Monitor, Lorapok Atlas
- Firefox AMO: Lorapok Add-ons β Atlas API Directory, XSnap Media Downloader, SubtitleMaster
- npm Registry:
@lorapok-labs/reportkit-ui,lorapok-atlas,lorapok-atlas-mcp,roast-api,lorapok-player - PyPI:
roast-api - Packagist:
lorapok/laravel-execution-monitor,maizied/roast-api - Snapcraft:
lorapokmediaplayer - Product Hunt: Lorapok Atlas Launch
# Clone the repository
git clone https://github.com/Lorapok/lorapok.github.io.git
cd lorapok.github.io
# Install client and agent dependencies
npm install
cd agent && npm install && cd ..
# Start local sensory development server
npm run dev
# Run type checking across client application
npx tsc --noEmit
# Compile agent TypeScript modules to CommonJS
npm run agent:build
# Execute full production build with static HTML SEO pre-rendering
npm run build
# Run LoLaBo autonomous publisher manually (single dispatch)
npm run agent:dispatch
# Start LoLaBo autonomous daemon service
npm run service:lolabo:daemonIn accordance with Loragent Governance Standards:
- Plaintext secrets, private keys, and passphrases are never committed to Git, passed via argv, or exposed in client bundles.
- Production credentials are synchronised through the centralized machine AES-256 vault:
cred get cloudflare account_id_lorapok cred get google_ai gemini_api_keys
Designed, engineered, and maintained by Mohammad Maizied Hasan Majumder (@Maijied) and the Lorapok Labs open-source collective.
Released under the MIT License.