What problem does it solve?
Building agents that run recurring internal tasks (scheduled posts, data syncs, webhook reactions) is risky without guardrails: they double-post, retry forever on expired credentials, or execute destructive actions without confirmation. This playbook produces ops automation agent prompts with idempotency, confirmation boundaries, and receipt-based output baked in.
Core Features & Use Cases
- System prompt template: Generates a complete agent prompt with trigger handling, idempotency key derivation, blast-radius caps, and a mandatory 7-field receipt output format.
- Safety boundary enforcement: Distinguishes autonomous reversible actions from destructive, financial, or high-blast-radius operations that must dry-run and stop for confirmation.
- Naming and capability guidance: Provides naming patterns, tool-selection rules (one integration per agent, scheduler for cadence), and anti-patterns to avoid.
- Use Case: A user asks for an agent that posts a standup template in #engineering every weekday at 9am; the playbook produces a "Daily Standup Poster" agent with a date-based idempotency key, Slack integration, scheduler workflow, and receipt-style reporting.
Quick Start
Ask the agent builder to create an agent that syncs new Stripe payments into a Slack channel every hour, using this playbook to enforce idempotency and receipt output.
