Live tool · Open source
A chain-agnostic EVM debugger I built and shipped. Paste a failing transaction and it tells you why it failed, proves the fix without touching the chain, and can broadcast the corrected call through your own wallet. Live, public, MIT.
The problem
When a transaction reverts, the chain gives you almost nothing: a status flag and, if you're lucky, a few bytes of error data. The engineer on the other end is usually mid-incident, guessing between a wrong address, a missing approval, a bad amount, or a contract that changed. I spent four years doing this decode work by hand on customer incidents at Fireblocks: pulling calldata apart, checking allowances, and re-simulating until the story held up.
The troubleshooter automates that exact workflow. Not a block explorer, not a wallet: a debugger built around the question "why did this fail, and what is the smallest change that makes it succeed?"
The approach
Runs the call on-chain with eth_call, no gas spent and nothing broadcast, to reproduce the failure exactly as the chain sees it.
Resolves the ABI, decodes calldata and the revert reason, and turns raw bytes into a named function call with typed arguments.
Re-simulates with targeted state overrides (balances, allowances, storage slots) to prove which missing condition causes the revert.
Renders the full call trace tree with asset diffs, so the failure point and its side effects are visible at a glance.
Can broadcast a corrected transaction client-side through a connected wallet, with shareable permalinks and a markdown report export.
Input is auto-detected: transaction hash, raw calldata, signed raw transaction, json call request. Ethereum by default, the top 10 EVM chains built in, or bring your own RPC.
Proof, not promises
These are unedited captures from the live tool, and each one links to a permalink that reruns the exact simulation. A USDC transferFrom with no allowance: the tool decodes the revert, names the missing condition, then proves the fix by overriding the allowance storage slot and re-simulating. Same call, no allowance, then allowance granted. Reverted becomes Success without spending a wei of gas.
Click a capture to rerun that exact simulation in the live tool. The override slot was verified against Ethereum mainnet before publishing.
How it's built
All simulation, decoding, and override logic lives in a framework-agnostic core package over a viem client, so the UI is a thin layer and the logic is unit-testable in isolation. A small Hono proxy fronts the RPC endpoints with an allowlist, rate limiting, and caching. Signing never leaves the browser: there are no server-side keys anywhere in the system.
ABI resolution falls back gracefully: your pasted ABI, then Sourcify and Etherscan, then a public signature database. Container calls like multicalls and Safe transactions expand into readable sub-calls.
Tested with vitest for the core, Playwright for the UI, and forked-Anvil tests so the simulation logic is exercised against real chain state, not mocks.
What it demonstrates
Revert selectors, state overrides, call traces, and asset accounting are the parts of EVM debugging that only make sense if you have actually done the work. The tool automates the exact decode-and-probe pattern I used on customer incidents at Fireblocks.
The input box accepts whatever a stressed integrator pastes into it: a hash, calldata, a signed blob, or a JSON call. Auto-detection, safe simulation before any broadcast, and a shareable permalink are all decisions about the person on the other end, not the technology.
Strict TypeScript, no server-side keys, client-side signing only, tested against a local Anvil fork with vitest and Playwright. Open source under MIT so the work can be read, not just believed.