You should not have to ask. Your agent should just tell. Today we are publishing Synorb Skills: seven jobs you set up once. After that, your agent runs each one on your Synorb Beacons on its own schedule and reports back, whether that is the Monday newsletter, the Friday memo or the morning episode on the way to work.
A skill is a short folder with a SKILL.md file, in the open Agent Skills format. It tells your agent which Synorb calls to make and how to write the result. It holds plain instructions and no hidden code, so you can read every line before you install it.
The seven skills
- Synorb Setup: connect in a minute, see a first cited briefing, and save the watches you care about.
- Newsletter Writer: a weekly issue in your own voice, every item cited to its source.
- Research Associate: a cited memo from any Beacon, with a timeline and confidence-labelled Claims.
- Podcast Producer: a two-host podcast script with voice directions, or a 60 to 90 second spoken briefing.
- Launch Radar: what AI companies shipped this week, from their own announcements.
- Thesis Tracker: what VC firms publish this month, mapped to your own thesis.
- Podcast Digest: the arguments from this week's business and tech podcasts.
Every skill runs on a Beacon
Each skill runs on a Beacon, a saved watch over many Streams: AI Companies (250+ Streams), Venture Capital (140+), Private Companies (1,000+) and Business & Technology Podcasts (120+), or a Beacon you save. Beacons carry a steady volume, so a skill always has material to work with. Every item is the publisher's own post, filing or episode, broken into Claims with confidence labels and a link back. The skills keep those labels, say whose words they are, and never add a fact the source does not hold.
Install once, use anywhere
In a coding agent, one command installs all seven:
npx skills add Synorb/skillsAny agent that can read a link can follow a skill from its page, for example synorb.com/skills/newsletter-writer.md. You can also download a skill as a ZIP. The gallery shows install steps for personal agents, chat assistants, coding agents, workflow tools, computer agents and voice agents, and a real result for every skill. The source is on GitHub under the MIT license.
We tested every skill end to end against live Synorb data before we published it, and the results on each page are what those runs produced. Synorb makes no model call when it serves a request. Your own model writes the newsletter, the memo or the script.