myroSecurity
Action receipts & audit
An autonomous agent you can't audit is a liability. Myro records every consequential action it takes — and, crucially, why it took it — in a tamper-evident ledger you can read in plain language.
Every action, with its cause
When Myro does something in the world — searches the web, reads an email, sends a message — it writes a receipt: what it did, the exact human instruction that triggered it, the data involved, the result, and the time. Nothing consequential happens without a trace.
Tamper-evident by design
Receipts are hash-chained: each entry cryptographically includes the fingerprint of the one before it. Altering or deleting any past record breaks the chain, and the break is detectable. The log can be appended to, but not quietly rewritten.
Provenance, not just logs
Standard logs tell you what happened. Myro's receipts tell you why — which instruction, which data, which standing authorization led to each action. An action with no legitimate human lineage is denied before it runs, and the denial is itself recorded.
Readable by anyone
The Activity view renders the ledger as plain sentences — “Sent an email to…”, “You approved…”, “Blocked an unauthorized action” — with the raw record one click away. It reads like an account of work done, not a developer's event stream.
Why it matters
You can hand Myro real work and still answer, at any time, exactly what it did and why — with a record that can't be quietly altered. That is the difference between automation you hope is behaving and automation you can prove was.