Skillget

schedx

agent
Security Audit
Fail
Health Warn
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 5 GitHub stars
Code Fail
  • rm -rf — Recursive force deletion command in install.sh
Permissions Pass
  • Permissions — No dangerous permissions requested
Purpose
This agent is a local-first CLI tool designed to schedule and manage recurring jobs. It allows users and AI agents to easily schedule shell commands, AI prompts, and HTTP webhooks using flexible time intervals.

Security Assessment
The overall risk is Medium. The tool inherently executes arbitrary shell commands via its scheduling features, which means any scheduled action should be reviewed carefully. It also makes outbound network requests to execute webhooks. No hardcoded secrets or dangerous permissions were found in the application code itself. However, the automated installation script (`install.sh`) contains a recursive force deletion command (`rm -rf`), which is a common but potentially risky pattern. If this script is altered or fails, it could lead to accidental data loss.

Quality Assessment
The project is quite new and has low community visibility with only 5 GitHub stars, meaning it has not been broadly tested or audited by a large audience. Despite the low visibility, the codebase is actively maintained with recent repository updates. It is properly licensed under the standard MIT license. Being written in Rust provides strong memory safety guarantees, and the tool includes automated continuous integration (CI) testing.

Verdict
Use with caution: The Rust-based tool is actively maintained and fit for its purpose, but low community adoption and the presence of `rm -rf` in the installation script warrant careful review before integrating it into sensitive environments.
SUMMARY

Scheduler CLI for recurring jobs, agent prompts, and webhooks

README.md

schedx

CI
crates.io
License: MIT

Scheduler CLI for recurring jobs, agent prompts, and webhooks.

One tool to schedule everything -- cron jobs, one-shot tasks, webhook calls, and AI agent prompts. Local-first, file-based, and designed to be used by both humans and agents.

Why I Built This

AI agents are great at doing things right now. But most real work happens over time -- a security audit that runs every night, a news digest compiled every morning, a deploy pipeline that checks back in 30 minutes. There was no clean way to give agents the time dimension. cron wasn't built for this. launchd wasn't built for this. So I built schedx.

I've been using it daily for a few months now. My Claude Code agent schedules its own security reviews. It compiles a morning news briefing from diverse sources and emails me a fair summary. I have a job that wakes up both Codex and Claude Code to run a full network audit on my mini PC, then passes both reports to a judge agent that only pings me if something's actually wrong. It's like hiring a sysadmin who never sleeps.

The idea is simple: if your agent can run a command, it can schedule one. And once scheduling is just another CLI call with JSON output, agents can build surprisingly sophisticated workflows across time -- without any glue code.

Install

From source:

cargo install schedx --locked

From GitHub Releases (macOS / Linux):

curl -fsSL https://raw.githubusercontent.com/Alireza29675/schedx/main/install.sh | sh

Quick Start

# Run something every hour
schedx add "every 1h" --run "echo hello"

# Remind me in 30 minutes
schedx add "in 30m" --run "say 'break time'"

# Post to Slack every weekday morning
schedx add "0 9 * * 1-5" \
  --webhook https://hooks.slack.com/services/T00/B00/xxx \
  --method POST \
  --header "Content-Type: application/json" \
  --body '{"text":"Good morning"}'

# Ask an AI agent for a weekly PR summary
schedx add "0 16 * * 5" \
  --prompt "Summarize this week's open pull requests"

# Manage your jobs
schedx list
schedx history
schedx logs <job-id>
schedx pause <job-id>
schedx resume <job-id>
schedx rm <job-id>

What It Does

Three action types, one interface. Schedule shell commands (--run), AI agent prompts (--prompt), and HTTP webhooks (--webhook) using the same CLI.

Flexible scheduling. Cron expressions, human intervals (every 5m, every 2h), one-shot timers (in 30m, in 2h), and exact ISO-8601 timestamps.

Full job lifecycle. Add, pause, resume, skip, and remove jobs. View run history, logs, and status -- all from the terminal.

JSON output. Every command supports --json for structured output. Built to be scripted by CI pipelines, shell scripts, and AI agents.

Local-first. All state lives as plain files under ~/.schedx/. No cloud, no daemon, no database. Atomic writes with fsync ensure nothing gets corrupted.

System-native scheduling. Integrates with systemd (Linux) and launchd (macOS) for the heartbeat tick. All scheduling decisions happen in schedx itself.

Real-World Agent Workflows

# Morning news briefing -- agent reads diverse sources, compiles a fair summary, emails you
schedx add "0 7 * * 1-5" \
  --prompt "Read top stories from Reuters, AP, Al Jazeera, and Ars Technica. Write a balanced 5-minute briefing and email it to me."

# Nightly security review -- two agents audit, a third judges
schedx add "0 2 * * *" \
  --run "claude -p 'Run a full security scan of this machine: open ports, failed logins, unusual processes' > /tmp/audit-claude.md && \
         codex -p 'Review network connections and firewall rules on this host' > /tmp/audit-codex.md && \
         claude -p 'You are a security judge. Review these two audit reports and only alert me if something needs attention: $(cat /tmp/audit-claude.md /tmp/audit-codex.md)'"

# Deploy canary check -- verify health 30 minutes after deploy
schedx add "in 30m" \
  --prompt "Check the /health endpoint and error rates for the last 30 minutes. Did the deploy go clean?"

# Weekly dependency audit
schedx add "0 10 * * 1" \
  --run "cargo audit && npm audit" \
  --name "dep-audit" --tag security

How It Compares

cron launchd at schedx
Recurring jobs Cron expressions only Cron + calendar intervals No Cron, every Xm/h/d, intervals
One-shot tasks @reboot only No native "run once at time X" Yes (single run at a specific time) Yes (in 30m, ISO-8601)
Webhooks Manual (wrap curl) Manual (wrap curl) Manual (wrap curl) Built-in (--webhook)
Agent prompts No No No Built-in (--prompt)
Job history No Last exit status only (launchctl list) No Append-only run history per job
Log capture Mailed or redirected manually stdout/stderr to file (configured per plist) Mailed Automatic per-run log files
Pause / resume / skip Remove and re-add crontab line launchctl enable / disable (no skip) No schedx pause, resume, skip
JSON output No No (launchctl outputs plists/text) No --json on every command
Agent-usable Agents must parse crontab text, no structured feedback Agents must generate XML plists, parse unstructured output Agents must parse queue text Agents get structured JSON I/O, --prompt as first-class action
Cross-platform Linux, macOS, BSDs macOS only Linux, macOS, BSDs Linux, macOS
Config format Crontab lines XML plists Interactive or piped stdin CLI flags, JSON state files

A note on agents: When an agent has to juggle crontab syntax, launchctl plists, and at queues just to schedule work, that's a lot of scattered context competing for the model's attention. I built schedx because I wanted one tool that brings the time dimension to the agent experience without the noise. It's a single CLI with JSON in and out, prompt scheduling as a first-class action, and queryable history. Reliable and intuitive for agents out of the box.

More Examples

See docs/EXAMPLES.md for real-world recipes covering DevOps, webhooks, AI agent workflows, one-shot tasks, monitoring, and job lifecycle management.

Security

  • Owner-only permissions on all data files and directories
  • HTTPS required for webhooks by default (http:// must be explicitly allowed)
  • Secrets and sensitive headers are redacted in output
  • No shell interpretation unless you pass --shell explicitly

Full policy in SECURITY.md.

Update

# Cargo installs
cargo install schedx --locked --force

# Direct installs
curl -fsSL https://raw.githubusercontent.com/Alireza29675/schedx/main/install.sh | sh

No silent background updates. Upgrades are always explicit.

Status

Pre-1.0. Semantic versioning applies, but breaking changes can still happen in minor releases while the CLI and persistence model settle. Linux and macOS only.

Contributing

License

MIT

Reviews (0)

No results found