agentleFS
notquenotque/vexjoy-agent/skills/infrastructure/headless-cron-creator/SKILL.md

headless-cron-creator

Skill448 starsChanged 5 days agoMITView on GitHub

Generate headless Claude Code cron jobs with safety.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 5 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Gate: All required parameters extracted.

    line 59
  2. Very specific:

    Create the wrapper script using crontab-manager.py generate-wrapper.

    line 63
  3. Very specific:

    Do not use the CronCreate tool -- it is session-scoped (dies when the session ends, auto-expires after 7 days).

    line 86
  4. Distinctive:

    Verify the generated script meets cron best practices.

    line 92
  5. Distinctive:

    Run the script in dry-run mode:

    line 94
  6. Distinctive:

    Gate: All checks pass. Proceed to Phase 4.

    line 107
  7. Distinctive:

    Ask the user for confirmation: before installing. Never install without explicit approval.

    line 122
  8. Very specific:

    If the path changes, regenerate the wrapper with --force.

    line 167

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Headless Cron Creator Skill

Generate headless Claude Code cron jobs from a task description and schedule. Creates a wrapper script with safety mechanisms (lockfile, budget cap, dry-run default, logging) and installs crontab entries. All crontab mutations go through scripts/crontab-manager.py, which writes to temp files and creates timestamped backups in ~/.claude/crontab-backups/ before every change -- never pipe directly to crontab - because a mid-stream pipe failure wipes the entire crontab.

Instructions

Phase 1: PARSE

Extract job parameters from the user's request.

Required parameters:

  • name -- short kebab-case identifier (e.g., reddit-automod, feed-health-check)
  • prompt -- what Claude should do each run (natural language)
  • schedule -- cron expression or human-readable interval

Optional parameters (with defaults):

  • workdir -- where to cd before running (default: current repo root)
  • budget -- max USD per run (default: 2.00; user may override)
  • allowed-tools -- which tools the headless session can use (default: Bash Read; user may override)
  • logdir -- where to store logs (default: {workdir}/cron-logs/{name})

Human-readable schedule conversion -- use off-minutes (7, 23, 47) instead of :00/:30 because every cron job on the system fires at round minutes, creating load spikes:

Human InputCron Expression
every 12 hours7 */12 * * *
twice daily7 8,20 * * *
hourly23 * * * *
daily at 6am7 6 * * *
weekly on sunday7 9 * * 0
every 30 minutes*/30 * * * *

Gate: All required parameters extracted. Proceed to Phase 2.

Phase 2: GENERATE

Create the wrapper script using crontab-manager.py generate-wrapper. Embed the prompt via bash heredoc to avoid escaping issues.

python3 ~/.claude/scripts/crontab-manager.py generate-wrapper \
  --name "{name}" \
  --prompt "{prompt}" \
  --schedule "{schedule}" \
  --workdir "{workdir}" \
  --budget "{budget}" \
  --allowed-tools "{allowed_tools}"

Review the generated script. Verify it contains:

  • set -euo pipefail
  • flock lockfile -- prevents concurrent runs of the same job
  • --permission-mode auto -- never use --dangerously-skip-permissions (auto is sufficient) or --bare (breaks OAuth/keychain auth)
  • --max-budget-usd -- caps spend per run (default 2.00 USD)
  • --no-session-persistence
  • --allowedTools
  • tee to per-run timestamped log file
  • Dry-run/execute toggle -- scripts do nothing destructive without --execute
  • Exit code propagation via PIPESTATUS[0]

Do not use the CronCreate tool -- it is session-scoped (dies when the session ends, auto-expires after 7 days). Use system crontab via crontab-manager.py instead.

Gate: Script generated and reviewed. Proceed to Phase 3.

Phase 3: VALIDATE

Verify the generated script meets cron best practices.

  1. Run the script in dry-run mode:
    bash -n scripts/{name}-cron.sh  # syntax check
  2. Run cron-job-auditor checks:
    • Error handling (set -e)
    • Lock file (flock)
    • Logging (tee, LOG_DIR)
    • Working directory (absolute cd)
    • PATH awareness (absolute path to claude -- cron has minimal PATH, so all commands must use absolute paths)
    • Cleanup on exit (lock release)

Gate: All checks pass. Proceed to Phase 4.

Phase 4: INSTALL

Install the crontab entry. Every entry gets a # claude-cron: <tag> marker so crontab-manager.py can identify and manage only its own entries without touching non-Claude crontab lines. All paths in the crontab entry must be absolute because cron has minimal PATH.

  1. Show the proposed entry:
    python3 ~/.claude/scripts/crontab-manager.py add \
      --tag "{name}" \
      --schedule "{schedule}" \
      --command "{absolute_script_path} --execute >> {logdir}/cron.log 2>&1" \
      --dry-run
  2. Ask the user for confirmation before installing. Never install without explicit approval.
  3. If confirmed:
    python3 ~/.claude/scripts/crontab-manager.py add \
      --tag "{name}" \
      --schedule "{schedule}" \
      --command "{absolute_script_path} --execute >> {logdir}/cron.log 2>&1"
  4. Verify:
    python3 ~/.claude/scripts/crontab-manager.py verify --tag "{name}"

Gate: Entry installed and verified. Proceed to Phase 5.

Phase 5: REPORT

Summarize the created cron job and print the exact commands to test and manage it.

Output:

  • Script path
  • Cron schedule (human-readable + expression)
  • Log directory
  • Budget per run
  • Tag for management
  • Commands for future management:
    python3 ~/.claude/scripts/crontab-manager.py list          # see all claude cron jobs
    python3 ~/.claude/scripts/crontab-manager.py verify --tag {name}  # health check
    python3 ~/.claude/scripts/crontab-manager.py remove --tag {name}   # uninstall

To modify an existing wrapper script, regenerate it with --force rather than editing in place.

Error Handling

Error: "tag already exists"

Cause: A cron entry with this tag is already installed. Solution: Either remove --tag {name} first, or choose a different name.

Error: "claude: command not found" in cron

Cause: Cron has minimal PATH; claude isn't in it. Solution: generate-wrapper resolves the absolute path to claude at generation time. If the path changes, regenerate the wrapper with --force.

Error: "crontab install failed"

Cause: System crontab service issue. Solution: Check crontab -l manually. Restore from ~/.claude/crontab-backups/.

References

  • scripts/crontab-manager.py -- all crontab mutations (add, remove, list, verify, generate-wrapper)
  • skills/infrastructure/deploy/references/cron-automation.md -- validation checks for generated scripts

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.