What's new
Every release of TaskBeacon, newest first. Version numbers stay below 1.0 for now. Atom feed
0.9.4 Ask Claude or ChatGPT about your jobs
Connect TaskBeacon to your AI assistant: ask what failed and why, and let it set up monitoring.
New
- Connect Claude, ChatGPT or another AI app (Settings → AI assistants) and ask what failed and why.
- Pro and Team: it can also create jobs, pause or resume them and change schedules. Nothing can be deleted.
- "Monitor all my scheduled scripts": it sets up the jobs, adds the lines (or a wrapper) and checks them.
- Ask which jobs were flaky, what didn't run while a machine was off, or for a better grace time.
- Claude plugin and script snippets, open source on GitHub.
Improved
- Jobs and the job page show recent runs as small bars.
- Docs, use case and comparison pages: the side menu keeps its place when you click.
0.9.3 A new look and clearer pages
Calmer design, a summary at the top of Jobs, and projects and machines as filters.
New
- New look and logo: calm greys, one accent colour, light and dark theme.
- Jobs: a summary at the top (needs attention, healthy, next run, latest change).
- Projects and machines are two filters that work together, each with its own … menu (rename, share, move jobs, delete).
- Job page: status, last ping, next run and the last 30 days at a glance.
- History: runs and problems by machine or project, and the longest runs.
- The once-a-day alert email is a short, clear summary.
- Code tab: Copy prompt for your AI, to let ChatGPT, Claude or Copilot add the lines to your script.
- Upgrade to Pro or Team in Settings → Plan & billing.
Improved
- Small design and layout fixes, especially on phones.
- Breadcrumbs on every page; the header stays on top while you scroll.
- New use cases: monitoring AI jobs and agents.
- New jobs start without rules; the row rules are added when a script first sends rows. Add a rule with + Add rule.
- Jobs and job pages refresh in place, without jumping to the top or closing what you opened.
0.9.2 No false Missed after our own outage
When TaskBeacon itself is unavailable, your jobs are not blamed for it.
Improved
- Runs that fall in a TaskBeacon outage are not marked Missed.
- After an outage longer than 3 hours you get one email about it.
0.9.1 Test alerts and safer sign-up
Fixes and improvements after an independent review of the beta.
New
- Send a test alert from Settings → Alerts.
- Code tab: choose whether a failure also sends the last lines of output.
- Stata code works on macOS and Linux; the PowerShell wrapper runs a .ps1 directly.
Improved
- Safer email confirmation, and sign-up works even when the email can't be sent right away.
- Fairer % OK, and no false Late or Missed for new jobs or changed schedules.
- Faster pages for accounts with many jobs, and clearer Docs for every language.
- Small design fixes.
Fixed
- Small fixes in the R, cron and SAS code, and on the error pages.
0.9.0 Public beta
The first public version of TaskBeacon. Everything below is what it does today.
Monitoring
- Your script pings when it starts and ends; a failure brings its exit code and the last lines of its output.
- Code to paste for Python, R, Stata, SAS, SPSS, PowerShell, Batch, cron, SQL Server Agent, Excel VBA and Node.js.
- Schedules like Windows Task Scheduler: daily, on chosen weekdays or dates, repeated every few minutes. A job is Missed or Late when a ping doesn't come in time.
- Website and API checks, with warnings before an SSL certificate or a domain expires.
- Integrations: Power BI, SharePoint/OneDrive, dbt Cloud, Fivetran and Azure Data Factory runs show up as jobs, without changing any script.
Data checks
- Row counts: an alert when a run loads far fewer rows than usual, or none.
- Freshness: an alert when the latest date in the data is too old.
- Your own numbers, with rules such as "at least" or "no more than".
- Keywords in the output: fail a run when its output contains a word, or accept it only when it does.
Alerts
- Email, Slack, Microsoft Teams, Discord, Google Chat, ntfy and webhooks.
- Reminders while a problem lasts, or everything in one email once a day.
- One "machine seems down" alert when every job on a machine is missed.
Reports & sharing
- Daily, weekly or monthly report by email.
- History for any period, with a ring chart of how runs ended and the jobs with the most problems.
- Status pages: a read-only link for a project, with 90 days of history per job.
- Projects shared with colleagues who can view or edit.
Plans
- Free, Pro and Team plans. Every paid feature is free for everyone during the beta.