Skills — a dedicated home for the tasks your AI agents run

Skills now have a real home in Simple Product: a new sidebar item, a dedicated /skills view, and a Create dialog shaped for the way skills actually get written.

What is a skill?

A skill is a markdown doc that describes a task you want an AI agent to perform on your workspace. Examples:

Every morning, read the last 24 hours of feedback and draft email outreach for anything mentioning feature X
When a card moves to Done, draft release notes from the linked work
Categorize incoming feedback into feature requests / bugs / questions

The skill itself is the artifact. The runtime that actually executes it is the next chunk of work — but you can write the skill now and it'll run as the runtime comes online.

What's shipped today

Skills sidebar item — 💡 icon, sits between Releases and Customers
/skills view — list of every skill in your workspace, search, cards/list layouts
Create dialog tuned for skills — placeholder text asks for a name like "Feedback analyzer" or "Spec writer," not a version number; blank doc by default so you can describe the task however you like
Skill docs hidden from /docs — they live in /skills only, so your general Docs view stays focused on actual reference material

What's coming

The skills runtime — what actually executes the markdown — is next:

Cron triggers — schedule a skill ("every morning at 8am")
Event triggers — bind a skill to a webhook event ("on feedback.created, run this skill")
Two execution paths — SP runs it natively for tasks fully inside SP's tool surface, or webhook out to your own runtime (Claude Code, Cursor, your own LLM service) for skills that need tools SP doesn't have access to (local files, IDE, custom MCPs)
Execution receipts — every run writes back to SP as a comment or linked doc, so you can see what happened, what tools were called, and what credits got spent

Write skills now; they'll start running as the runtime lands. No migration, no porting — the skills you author today are the same artifacts the future runtime will pick up.

Suggested format

There's no spec yet, but skills read best when they're structured like a clear prompt:

## What this does
One paragraph plain-English description.

## When to run
- Trigger conditions (event, cron, manual)

## Inputs available
- Data the agent should pull (cards, docs, feedback, etc.)

## Steps
1. Numbered, ordered actions

## Output
What gets produced and where it lands (a draft doc, a comment, an email draft, ...)

## Constraints
What NOT to do. (E.g. "always leave as draft, never auto-publish")

We'll publish a fuller skills spec alongside the runtime.

Related: webhooks shipped the same day

Skills are one half of Simple Product's automation story. Webhooks — also shipped today — are the other half: events your own services can react to. Together with cron (coming), they're the three primitives we'll build all automation on.

What we'd love to hear

Which trigger should we ship first — cron or events? (Cron is simpler. Events feel more useful for reactive workflows.)
What skills do you want to write first? The most-requested ones are the most likely to shape how the runtime works.

Send feedback in-app or drop a note in the Skills view directly.