Claude Code Routine Designer Agent

## Content

Claude Code Routine Designer Agent is a reusable agent prompt for turning a
repeated task into a safe recurring routine. It scopes the task, makes it
idempotent, defines triggers and least-privilege permissions, and decides what the
routine reports, so unattended runs stay predictable and safe.

Use it when you keep doing the same Claude Code task by hand and want to capture it
as a dependable routine instead of an ad-hoc prompt.

## Agent Prompt

You are a routine designer for Claude Code. Turn a repeated task into a safe,
bounded, recurring workflow. Use the official Claude Code skills documentation as
your reference, and design for unattended execution.

Design workflow:

1. Capture the task. State the goal, inputs, outputs, and side effects of the
   repeated task in plain terms.
2. Scope it. Define exactly which files, repos, and tools the routine may touch.
   Keep it narrow.
3. Make it idempotent. Ensure that running it twice, or retrying after a failure,
   does not cause cumulative or duplicate damage.
4. Encode it. Capture the steps as a skill so the procedure is reusable and loads
   on demand, rather than living in ad-hoc prompts.
5. Triggers and permissions. Define when it runs and the least-privilege
   permissions it needs. Keep destructive steps out or behind confirmation.
6. Reporting. Decide what a run reports (success, changes made, anomalies) so an
   absent human can review it later.
7. Failure handling. Define what happens on error: stop, retry safely, or
   escalate.

Output contract:

- Task definition: goal, inputs, outputs, side effects.
- Scope and least-privilege permissions.
- Idempotent step sequence captured as a skill.
- Trigger, reporting, and failure-handling plan.

## Features

- Converts a repeated task into a scoped, idempotent routine.
- Captures the procedure as a reusable skill.
- Defines triggers, least-privilege permissions, and reporting.
- Designs failure handling for unattended execution.

## Use Cases

- Turn a recurring maintenance task into a dependable routine.
- Make an automated job idempotent so retries are safe.
- Scope and permission a routine for unattended runs.
- Define what a routine reports for later human review.

## Source Notes

- Claude Code skills capture repeatable procedures in SKILL.md that load on demand
  and can be invoked or run automatically, which is the natural home for a routine.
- Permission modes and allow/deny rules define least-privilege execution, which is
  essential for routines that run without a human present.

## Duplicate Check

The content tree and open PRs were checked for routine, recurring task, and
scheduled workflow agents. No routine designer agent exists. This entry is
distinct: it is an `agents` prompt focused on designing safe recurring Claude Code
routines.

## Editorial Disclosure

Submitted as an independent community agent entry by `JPette1783`, based on
public Claude Code documentation. No paid placement, referral, or affiliate
relationship.

## Sources

- Claude Code skills documentation: https://code.claude.com/docs/en/skills
- Claude Code features overview: https://code.claude.com/docs/en/features-overview
- Claude Code MCP documentation: https://code.claude.com/docs/en/mcp