Receipt diff · structural
Compare two runs. See what changed.
Two signed sessions, folded to two different roots. This is a structural diff of what each run did, agent by agent, file by file. It catches the things a summary hides, like an agent that silently swapped its model.
deploy to stagingreceipt A
mroot_3f19a7c4e2b08d5a…
3 agents · 4 leaves
deploy to productionreceipt B
mroot_b108f8ef73065806…
4 agents · 6 leaves
model change detected
executorgpt-4o→claude-sonnet-4
metricABΔ
agents34+1
tool calls3254+22
tokens69,000101,800+32,800
files written35+2
artifacts46+2
merkle leaves46+2
handoffs24+2
agents
orchestratorclaude-opus-4→claude-opus-4tools +1
plannergpt-4o→gpt-4otools +1
executorgpt-4o→claude-sonnet-4tools +8CHANGED
tester·→kimi-k2NEW
files written
src/deploy/config.toml+8 -3MODIFIED
src/api/health.ts+24 -9MODIFIED
CHANGELOG.md+12 -0MODIFIED
Cargo.toml+1 -1NEW
tests/smoke/deploy.test.ts+47 -0NEW
This is a structural diff. Neither receipt's signatures are re-verified here; the two differ, so they fold to two different signed roots. Run treeship verify locally to check each one's authenticity.