From monitoring-jobs
Create a global scheduled Job by turning a user automation request into ~/.agents/jobs/<name>/JOB.md and optionally registering it with the jobs CLI.
How this skill is triggered — by the user, by Claude, or both
Slash command
/monitoring-jobs:job-creatorThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Create a schedulable global Job that matches the repository Jobs specification and the current `jobs` CLI.
Create a schedulable global Job that matches the repository Jobs specification and the current jobs CLI.
~/.agents/jobs/<job-name>/JOB.md.JOB.md with YAML frontmatter plus Markdown instructions.
name, cron, description.condition, allowedSkills, timeout, retry, tags.node cli/bin/openjob add ~/.agents/jobs/<job-name> only if the user wants the job registered now.~/.agents/jobs.json is missing and registration is requested, run node cli/bin/openjob init first.jobs add is the source of truth for parsing and cron validation.description: One sentence describing what runs and when it is useful.cron: Use standard 5-field cron. Prefer simple schedules.condition: Add when the job should skip unless a concrete state is present.allowedSkills: Restrict to the skills the job actually needs. Omit it when broad tool use is expected.timeout: Default to 60; increase for repo-wide implementation jobs or slow external work.retry: Default to 0; use 1 for transient network/API work.tags: Use 1-4 short tags for browsing and filtering.Common cron values:
| Intent | Cron |
|---|---|
| Daily at 9 AM | 0 9 * * * |
| Daily at midnight | 0 0 * * * |
| Every Monday at 9 AM | 0 9 * * 1 |
| Every 30 minutes | */30 * * * * |
| Every 6 hours | 0 */6 * * * |
| Monthly on the 1st at 9 AM | 0 9 1 * * |
Use this structure unless the task needs a stronger custom shape:
---
name: example-job
cron: 0 9 * * *
description: Check project state every morning and report actionable follow-ups.
condition: Check whether there is new project state to review.
allowedSkills: [jobs-scheduler]
timeout: 60
retry: 0
tags: [automation]
---
# Example Job
## Objective
Describe the outcome the agent should produce.
## Execution Steps
1. Inspect the current state relevant to the condition.
2. Execute the requested automation.
3. Verify the result before reporting completion.
## Output Requirements
Report what changed, what was verified, and any blockers.
## Notes
Keep side effects scoped to the task described in this job.
The current jobs CLI supports:
jobs init: create ~/.agents/jobs.json and ~/.agents/jobs/.jobs add <path>: parse JOB.md, copy it to ~/.agents/jobs/<name>/JOB.md, register the job globally, and sync enabled jobs to Claude scheduled tasks.jobs list: display registered jobs.jobs run <name>: print the prompt for manual execution.jobs enable|disable|remove <name>: update registry state and sync.jobs sync: sync enabled registry entries to Claude scheduled tasks.Prefer node cli/bin/openjob ... inside this repository unless openjob is known to be installed globally.
npx claudepluginhub homyeeking/openjob --plugin automation-jobsCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.