Authority Inheritance Across Temporal Scope is a structural anti-pattern in which an initial authorization is used to justify subsequent actions without re-evaluation at each irreversible execution boundary. The pattern is characterized by:
This pattern concerns the temporal structure of authorization, not the quality of initial approval.
- Authorization granted once that persists indefinitely unless explicitly revoked.
- No governance boundary check at each irreversible action within a task scope.
- No capability to halt execution when context, risk, or requirements change mid-task.
- Authority that accumulates across sub-tasks without proportional accountability.
The mechanism typically manifests through:
The mechanism treats authorization as a state that, once granted, covers all subsequent actions within the task scope. The scope boundary is typically implicit or unbounded.
- Persistent agents that receive one authorization at task initiation then operate autonomously for extended periods.
- Task decomposition where parent task authorization is implicitly assumed to extend to all sub-tasks.
- Multi-step workflows where intermediate governance gates are skipped in favor of initial approval.
- "Background agents" that execute over hours or days with a single up-front authorization.
Bounded authorization decisions produce unbounded exposure.
When authorization is inherited across temporal scope: • A user approves "send email to the team" and the agent sends 50 emails over 3 hours. • A user approves "update the codebase" and the agent modifies 200 files across 12 repositories. • A user approves "manage my calendar" and the agent declines meetings, reschedules commitments, and sends responses for weeks.
The initial authorization decision cannot anticipate all actions that will be taken under its scope. The user authorized an intent, but the system executed a sequence — and the sequence may diverge arbitrarily from intent.
Authorization scope becomes unknowable at decision time.
The structure creates a temporal mismatch: From: "I authorize action X at time T." To: "I authorize all actions the agent deems consistent with X, from T until task completion or explicit revocation."
The user cannot know what they are authorizing because the authorization covers future actions that depend on future context. The authorization decision is made with incomplete information about its own scope.
COUNTER-PATTERN: Irreversibility-Bounded Authorization
Authorization must be re-evaluated at each irreversible execution boundary. The governance layer must: • Classify actions by reversibility and require fresh authorization for irreversible effects. • Maintain temporal scope limits that require re-authorization after defined periods. • Support context-aware re-evaluation when task state diverges from initial authorization context. • Enable granular revocation that halts specific action classes without terminating entire tasks.
The user may authorize a goal. Each irreversible step toward that goal requires its own authorization at the moment of execution.
Authority Inheritance Across Temporal Scope converts bounded authorization decisions into unbounded execution authority. The resulting system cannot ensure that actions taken under inherited authority would have been authorized if evaluated independently at execution time.