12 live dashboards, one weekend, zero build steps

Last weekend we shipped 12 interactive dashboards. Not prototypes. Not wireframes. Live, publicly accessible, TLS-secured dashboards across every core kcolbchain project. Each one is a single index.html file with inline JavaScript. No webpack config. No React hydration. No 47-megabyte node_modules directory. Just HTML that works.

Here is how, and more importantly, here is why.

The stack that is not a stack

Every dashboard follows the same pattern: one HTML file, inline CSS custom properties that inherit from a shared design token set, and vanilla JavaScript that fetches data from public APIs or runs simulations client-side. Caddy reverse-proxies each subdomain and auto-provisions Let's Encrypt certificates. Deploy means scp index.html to the right directory. That is the entire CI/CD pipeline.

This is not laziness. This is a deliberate architectural choice. When your dashboard is a single file with zero dependencies, it loads in under 200ms on a cold cache. It never breaks because a transitive dependency pushed a bad patch. It works on a phone in Lagos the same way it works on a MacBook in San Francisco. And when you need to change something, you open one file, edit it, and push.

We have been building software for over a decade. The lesson that keeps compounding: the fastest code to debug is the code that does not exist.

The twelve

Here is every dashboard, what it does, and where to find it:

• gas-oracle — Live L2 gas prices and blob-fee prediction for Arbitrum, Optimism, Base, Scroll. Polls public RPCs every 12s. L1 blob base fee derived from excessBlobGas via EIP-4844's fake_exponential in-browser. Per-chain 10-block EMA forecast + sparkline. GitHub · Docs · Tutorial
• depeg-monitor — Real-time USDC / USDT / DAI peg status. Median-aggregates prices from Binance, Coinbase, and CoinGecko. Adjustable warn/critical thresholds (bps). Sparkline with peg + threshold reference lines. Alert log fires on severity transitions. GitHub · Docs
• token-simulator — Interactive tokenomics simulator with the Create Protocol V4 burn-toll preset. Adjustable supply, vesting, revenue streams, burn toll, vault yield, operating costs, growth tapering. Endogenous price via constant-product AMM. Consistency alerts flag when the model breaks. GitHub · Tutorial
• rwa-toolkit — Side-by-side ERC-3643 / ERC-1404 / KAIO comparison + live transfer-restriction simulator. 8 pre-seeded wallet personas, configurable token policy (jurisdictions, accreditation, max holders, lock-up, sanctions). Verdict with ERC-1404 restriction code. GitHub · Docs
• switchboard — Agent payment protocol explorer. Five tabs: x402 (Linux Foundation), MPP (Stripe/Tempo), AP2 (Google), Circle Nanopayments, and on-chain AgentEscrow. Click through 4-step SVG sequence diagrams with wire-level HTTP/JSON/Solidity. Cross-protocol compatibility matrix. GitHub · Docs · Tutorial
• erc721-ai — AI model tokenization gallery. 5 mock models (SDXL LoRA, Llama-3, Whisper, Stable Audio, CodeLlama). ERC-721 AI metadata schema. x402 per-inference metering calculator with HTTP 402 flow diagram. Provenance timeline. GitHub · Docs
• quoter — Market-making strategy backtester. Three strategies: Avellaneda-Stoikov, Grid, Inventory-Skew. Per-strategy parameter sliders, 500-tick Brownian-motion sim, mid/bid/ask + PnL plots, summary stats (Sharpe, drawdown, fill count). Side-by-side comparison mode. GitHub · Docs
• meridian — RWA market-making agent simulator. Switch between constant_spread and adaptive_spread across 4 asset presets (private credit, T-bill, real estate, long-tail). Live KPIs: PnL, inventory, fill rate, adverse selection, Sharpe. Order ladder + recent fills. GitHub · Docs · Tutorial
• stablecoin-toolkit — Stablecoin mint/redeem simulator with burn-toll impact chart at 3 volume tiers. Reserve composition donut (vault + yield + treasury). 4 compliance module toggles. 3 multi-geography deployment templates (US/EU/LATAM). GitHub · Docs
• monsoon — Airdrop farming research scaffold. 4 strategy cards (bridge, DEX, lending, LP). Target protocol table with scout fit scores. Wallet rotation simulator with scheduling timeline. Airdrop-rights NFT concept. GitHub · Docs
• audit-checklist — Vulnerability check explorer. One page per Foundry check (Reentrancy, AccessControl, Oracle, Upgrade, FlashLoan, Vault). Each page shows detection logic, required hooks, and the real VulnerableVault bug it flags with inline Solidity. GitHub · Docs · Tutorial
• superchain-trace — OP Superchain cross-chain message lifecycle simulator. Trace a message through: submitted, indexed by OP-Supervisor, safety levels (unsafe, safe, finalized), relayed on destination. Chain selector, animated trace output. GitHub · Docs

Why zero-build matters more than you think

The frontend ecosystem has spent a decade convincing everyone that you need a build step. That you need JSX compilation, tree-shaking, code-splitting, server-side rendering, and incremental static regeneration. For a marketing page. For a dashboard. For a form with three inputs.

We are not anti-tooling. We use Foundry for smart contracts. We use Python for data pipelines. We use the right tool for the job. But for dashboards that fetch data and render charts, the right tool is the platform itself: the browser.

Every abstraction layer you add is a layer that can break, a layer that needs updating, a layer that slows down your feedback loop. When we wanted to add a new chart to the gas oracle, the cycle was: edit the file, refresh the browser. Not: edit the file, wait for hot module replacement to maybe work, debug a stale cache, restart the dev server, discover that your version of the charting library conflicts with your version of the framework.

Shipping fast is not about typing fast. It is about removing the steps between your intention and the user seeing the result.

The infrastructure layer

Caddy deserves a shoutout here. Each dashboard lives in its own directory under the main site — kcolbchain.com/gas-oracle/, kcolbchain.com/meridian/, and so on. Caddy's try_files resolves the path to index.html automatically. Adding a 13th dashboard means creating a directory, dropping in an index.html, and pushing. No config changes.

We run this on a single VPS. Not a Kubernetes cluster. Not a fleet of serverless functions. One box, one reverse proxy, twelve directories. Monthly cost: less than a team lunch.

What comes next

These dashboards are not just demos. They are the interactive layer of our research. Every paper we publish, every protocol we analyse, every tokenomics model we design — it gets a dashboard. Because a static PDF is fine for peer review, but a live simulation is what changes minds.

We are already working on dashboard 13: a cross-chain bridge monitor that tracks lock/mint asymmetries in real time. If a bridge's TVL on one side diverges from the other by more than a configurable threshold, it lights up red. The kind of thing that would have saved people a lot of money in 2022.

All twelve dashboards are open source. The repos follow the pattern github.com/kcolbchain/{name}. Fork them, run them locally, break them, fix them, send a PR. That is the whole point.