skills.rest
alirezarezvani
alirezarezvaniCOMMUNITY
Explore all skills by this author →
Website𝕏@RezaRezvaniBlnGitHub

run-without-you

Create recurring POSIX-cron deployments for Claude Managed Agents with self-grading outcomes.

25.3k|3.6k|Updated Oct 19, 2025
Search Tags:#scheduling#Deployment#Automation#recurring jobs#CRON#Claude Managed Agents#timezone validation
One-click install
npx skills add https://github.com/alirezarezvani/claude-skills --skill run-without-you
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: run-without-you
Source: https://github.com/alirezarezvani/claude-skills/tree/main/agent-launcher/skills/run-without-you
Command: npx skills add https://github.com/alirezarezvani/claude-skills --skill run-without-you

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Turning a tested Claude Managed Agent into an unattended recurring job requires building a correct POST /v1/deployments payload, validating cron schedules and IANA timezones, and handling DST edge cases — all of which are error-prone when done by hand.

Core Features & Use Cases

  • Deployment Payload Builder: Generates the POST /v1/deployments JSON with initial_events (user.message plus an optional nested user.define_outcome so each firing self-grades) and prints the BYOK curl commands to create and test it.
  • Cron & Timezone Validation: Validates 5-field POSIX cron expressions and IANA timezone IDs, and prints the wall-clock DST note warning about skipped or double-fired times.
  • Roadmap Finalization: Writes NEXT-DIRECTIONS.md from the build sheet's deferrals, recording the versioned upgrade plan.
  • Use Case: You have a graded agent and want it to run a nightly audit at 9 AM Europe/Berlin — validate the schedule, build the deployment payload with a nested self-grading outcome, fire one manual test run, then leave the cron in place.

Quick Start

Ask the agent to put my graded agent on a schedule that runs every morning at 9 AM in Europe/Berlin and have each run grade itself.

Frequently Asked Questions about run-without-you

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I schedule a Claude Managed Agent to run on a cron cadence?▼

Build a POST /v1/deployments payload with deployment_builder.py, supplying a 5-field cron expression, IANA timezone, and an initial user.message event. Create it with the printed BYOK curl command, then fire one manual run to test before trusting the schedule.

How do I validate a cron expression and timezone before deploying?▼

Run cron_validator.py with --cron and --timezone flags. It checks the 5-field POSIX cron syntax and ranges, resolves the IANA timezone against the local zoneinfo database, and exits 1 on invalid input.

Does cron scheduling handle daylight saving time correctly?▼

CMA deployments use wall-clock DST semantics: a nonexistent local time on spring-forward is skipped, and a repeated local time on fall-back fires twice. Avoid scheduling between 02:00 and 03:00 in DST-observing zones if exactly-once execution matters.

Can each scheduled run grade its own outcome automatically?▼

Yes. Pass --nest-outcome to deployment_builder.py to include a user.define_outcome event alongside the user.message in initial_events, so every firing carries the rubric and self-grades its result.

Do the scripts call the Anthropic API directly?▼

No. All three tools are stdlib-only and make no network calls. deployment_builder.py only prints the BYOK curl commands; you create and test the deployment yourself using your own ANTHROPIC_API_KEY.

What are the limits and safety requirements for scheduled deployments?▼

An organization supports at most 1,000 deployments, and there is no spend cap inside CMA. Recommended rails include always_ask MCP, limited networking, read_only untrusted memory, max_iterations per firing, and a workspace spend limit.

Related Skills

Explore More Software Engineering Skills→
rhpdsrhpds

23-background-agents-goal-mode

Manage Claude Code background sessions with goal-based autonomous execution.

Official
Advanced
Elijah-JElijah-J

ar-closeout

Summarize perf autoresearch results, verify final state, and archive artifacts for handoff.

Community
Advanced
anajuliabitanajuliabit
6

Skill Evals

Validate Aeon skill outputs against an assertion manifest and detect regressions.

Community
Advanced
trohitgtrohitg
356

cron-scheduler-skill

Define cron-based schedules with timezone support for automated workflow runs.

Community
Intermediate