Receipts you can check
How korgex's tamper-evident record works — in plain terms.
Every run korgex does is written to a logbook where each entry is sealed to the entry before it. Change, delete, or reorder even one entry and the seal breaks — visibly, and at the exact spot. That's what lets you check an AI's work instead of trusting it.
See it for yourself
korgex verify # prove the recorded run wasn't altered (exit 0/1 — works in CI too)
korgex trace # the step-by-step trace: what it did, and what caused each step
korgex why src/auth.py # walk back from a changed file to the request that caused it
korgex cost # what the run cost, counted from the recorded tokensTamper-evident vs tamper-proof
Out of the box the record is tamper-evident — any edit is detected. Set a secret key
(KORG_LEDGER_HMAC_KEY) and it becomes tamper-proof: edits can't be faked at all.
Hand someone a receipt
You can mint a single portable file that proves what a run did — it checks offline, and you can open it in any browser to watch it re-verify itself:
korgex receipt --claim "shipped /healthz + passing test" --html receipt.html
korgex receipt verify receipt.korgreceipt.json # ✓ VALID / ✗ INVALIDGate it in CI
Drop the verify-ledger GitHub Action into any repo to fail the build if an agent's record
doesn't check out — zero trust in the tool that produced it.
Why this matters
Audits, compliance, debugging, or simple peace of mind: you get honest proof of what the AI actually did. As far as we know, no other coding assistant offers this.