Turn agent help requests into a capacity-aware interruption budget for deciding what interrupts now, what batches, what stays in shadow mode, and what stops.
This is a complete workflow library with 5 individual skills. Download the full library or pick the specific skill folder your team needs first.
Some AI tools expect one skill folder per upload. Download the full library when you want the whole workflow, or download an individual skill when you only need one job done.
Use when an agent request or proposed action may require immediate human interruption because the action is high-impact, irreversible, sensitive, customer-facing, credentialed, regulated, destructive, payment-related, production-affecting, or prompt-injected.
Use when agent uncertainty, repeated minor tool errors, low-risk classifications, or non-urgent decisions should be reviewed in a batch instead of interrupting a human immediately.
Use when a new, changed, or poorly calibrated agent workflow should simulate actions, log evidence, and collect human disagreement data before receiving write or send authority.
Use when an agent should stop instead of interrupting or continuing because the reviewer queue is overloaded, retry budget is exhausted, evidence is missing, objective is ambiguous, dependency is unhealthy, or the same failure repeats.
Use when approval, rejection, override, queue, shadow-mode, or reopened-case signals should update the Skill, guardrail, escalation threshold, or review schedule.
Use the public library when the workflow is low-risk, the inputs are already sanitized, and a team member can review the output before it reaches a buyer or customer.
Do deeper review when this workflow touches real tools, data sources, role ownership, approval paths, or customer-facing output.