Merkle Receipt Explorer
failedAn interactive ledger that gives every step of an agent run a cryptographic receipt, chains them into a Merkle root, and lets you tamper with one step to watch verification break in real time.
Approach
Single index.html with inline CSS and JS, no build step and no dependencies: the browser's native crypto.subtle.digest (SHA-256) computes every leaf and internal node hash client-side, so the one claim that matters — tampering is cryptographically detectable — is real, not simulated. An authored array of ~22 agent steps (tool name, input, output, timestamp) is the seed data; on load an async function hashes each step's JSON into a leaf, pairs leaves bottom-up into a Merkle tree, and renders the root plus a per-step table with each leaf hash shown. Each row has an inline 'edit' control that lets the visitor change that step's output text; on submit the leaf and every ancestor hash up to the root are recomputed and re-rendered, and a verification badge flips from valid to broken because the recomputed root no longer matches the original. A reset button restores the tampered step and the original root. Everything — crypto, tree construction, re-render — runs in the one HTML file with no network calls after load.
The source post
https://x.com/zamboMCP/status/2109415889347269108
Scoring
Pick
| surprise | 4 |
|---|---|
| demonstrability | 5 |
| self_containedness | 5 |
| honesty | 5 |
zambo.dev's live 100-step agent run with Merkle receipts is the interesting idea here, but the artifact worth building isn't a clone of their running agent — it's the verification mechanism itself, made clickable. An authored sample agent trace gets hashed into a Merkle tree entirely in-browser; editing any single step's receipt recomputes the root and visibly breaks the chain, so the one claim that matters — 'tampering is detectable' — is real cryptography running client-side, not a mockup. It needs no account, key, or live network call at build or run time, ships as one static page, and a stranger gets the point within one click of editing a step. ALTRIX's forecast explorer was the closest competitor on self-containedness but scores lower on surprise (slider-driven chart demos are a well-worn pattern) and carries honesty risk if a historical-data toy reads as a real predictive claim.
Review
no scores recorded
Cost
| total | $0.0000 |
|---|
What it looked at
Voxt open-source Mac voice input tool** (local SGLang Omni + Qwen3.8-27B refinement): runs entirely on-device with hotword + LLM correction for technical AI terms; demo video shows real-time accuracy on M5 Pro.
Live agent workflow with Merkle receipts** (zambo.dev): single agent executes 100 steps, mints cryptographically verifiable receipts rolled into one inspectable root—exact run viewable now at the provided link.
$7 ESP32 + local Qwen3.8-Flash controls stock iPhone**: no jailbreak/developer mode, everything LAN-only on 2× DGX Spark; open-source repo and video demo of the full hardware+AI loop.
SuryaSaathi solar-panel AI agent**: identifies output drops on real rooftop systems and guides free repairs; working demo linked in the post.
Resurface bookmark-to-reel + agent skills**: turns saved tutorials into watchable reels that become callable skills for Codex/Claude; live demo shown at the hackathon.
ALTRIX Forecast Lab crypto scenario explorer**: interactive web demo lets users run historical-data forecasts with adjustable horizons directly in the app.
Lain agent that versions its own tools**: builds and saves new tools persistently across runs so they become reusable personal infrastructure; phase-2 demo video.
Cat AI Unity demo** (game dev): capsule-based cat agents behave naturally in a horror-game environment with no human players; short playable-feeling clip.
Gate
| pass | project directory exists — /Users/artax/code/builds/2026-10-12/project |
|---|---|
| pass | no build step needed — static project |
| pass | build output with index.html — /Users/artax/code/builds/2026-10-12/project |
| pass | index.html is a document — 22600 bytes |
| pass | local asset references resolve |
| pass | page loads without console errors |
Notes
Omitted for lack of
None. All requested capabilities use native browser APIs and inline data.
Notes
publish: wrangler deploy failed: exit 1 ⛅️ wrangler 4.149.0 ──────────────────── [31m✘ [41;31m[[41;97mERROR[41;31m][0m [1mThe Pages project "merkle-receipt-explorer" does not exist.[0m Maybe you intended to deploy a Worker project instead? Workers are the recommended way to deploy all new projects. If so, run `wrangler deploy`. If you are targeting an existing Pages project, verify that the project name is correct and that it exists in your account (the account in use has id 057bfca49415a4fdfa15f9883ffd3052). Otherwise, if you are trying to create a new Pages project, start by running: `wrangler pages project create` (though we strongly recommend using Workers instead). 🪵 Logs were written to "/Users/artax/.wrangler/logs/wrangler-2026-10-12_08-08-08_439.log" — self-repaired in auryn@24f8f3b: The bug: `publish.deploy()` discarded the result of `wrangler pages project create` entirely, assuming any failure meant. Re-run to apply it; finished stages are reused.