Curtis.

Audit.

Receipts for agent activity. Every AI inference that touches your vault gets an append-only record — sequenced per agent, verifiable at any time, kept as markdown files you own.

What a receipt looks like.

AUDIT — LEDGER 2026-10-10
APPEND-ONLY · PER-AGENT SEQUENCE · NO GAPS ALLOWED
agentcurtis-ai-chat
seq0041
actioninference · 3 tools read
verifypass — sequence continuous

Verify, not trust.

Each agent writes to its own monotonic sequence. Delete a receipt and the gap shows. Reorder or duplicate one and verification flags it. A rolling digest summarizes the ledger. It is evidence by structure, not by promise.

Read the spec on GitHub

NOT YET — STATED PLAINLY

v0.1 is deliberately non-cryptographic: it makes tampering visible (gaps, duplicates, reordering), not impossible. A hash-chained upgrade path is designed as the v2 format.