Skills

Simple Product is your product brain — the place you already store every conversation with customers, every idea, every card, every release. Now, skills lets you automate that work. Write English instructions once; agents run them against the same context you already have.

Skills are natural-language docs that agents can run. Write what should happen — read this card, draft an email, create a release — and Simple Product will run it either automatically when something triggers it, or manually when you (or an external agent like Claude Code) invoke it.

What you can do

Write a skill in plain English. Create a doc of type skill, describe the workflow, save. That's it — the doc is the skill.

Two ways to use every skill:

Manual (default): the skill is a prompt template. External agents — Claude Code, Cursor, the in-app chat — can read the skill body and execute the steps themselves using Simple Product tools. Or you click Run Now on the skill doc.
Automated: click Automate and Simple Product wires the skill to its triggers. It fires on its own when the event happens — no more attention required from you.

Change your mind whenever. When you update the text of a skill that is automated, the skill will keep running based on the old instructions. That way, you can edit a skill without breaking it. When you're update is ready, the Update skill button re-reads the skill and updates the automation from your latest edits. Stop automating turns off the trigger listening without deleting the skill.

What agents can do inside a skill

Triggers (things that fire an automated skill):

Card moved between stages — "when a card lands in Done…"
Feedback received — "when a customer submits feedback…"
Doc created, updated, published, unpublished, deleted
Release published
Person or org added
Card created or updated

Actions (things a skill can do):

Read: cards, docs, feedback, linked people, board structures, full-text search across docs and cards
Create: cards on any board (by name), release docs, email drafts to specific people
Update: card titles and descriptions, move cards between stages
Link: docs to cards, cards to people
Save a run summary explaining what happened

All actions stay inside your workspace — no cross-workspace access, no external APIs (yet), no auto-sent emails (drafts only, for review).

Run history + transparency

Every skill run — automated or manual — records a full transcript. See it at Skills → Runs. Each row shows:

The trigger that fired it ("Feedback received — …", "Card moved to Done — …")
The agent's own summary of what it accomplished
Duration + token cost
Expandable full transcript with every tool call and result

Something surprising happened? The transcript is the answer.

Example skill bodies

Auto-triage incoming feedback

When new feedback is created:

Classify as a bug or feature request.

If a bug, search the SP Bugs board for a similar existing card.

If a feature, search the User Feedback Ideas board.

If a similar card exists, link the feedback to it. If not, create a new card and link the feedback.

Draft release + customer follow-ups when a card moves to Done

When a card moves to Done:

Read the card's details.

Find feedback docs linked to the card.

For each customer behind that feedback, draft a personal email thanking them and describing what shipped.

Create a release doc summarizing what shipped.

Save a one-sentence summary of the run.

Skills are prose, not code. Write like you're briefing a smart teammate.

Learn more

Full details for humans and AI agents:

Skills overview — the concept, states, lifecycle
Skill authoring guide — supported triggers, complete tool inventory, best practices, more example bodies

Both docs are public, so any external agent with your MCP can read them and know how to write and run skills correctly.

Coming next

External integrations — Slack, Gmail, GitHub — so skills can act outside SP
Inbound webhooks — Stripe, Vercel, anything with an HTTP POST — become skill triggers
Cron / scheduled triggers — "every Monday at 9am, do X"
Persistent memory for long-running skill lifecycles

The engine is designed for these — each one plugs into the existing event and tool registries without engine changes.

Try it now: create a skill doc, describe a workflow in plain English, click Run Now to test, then Automate when it's doing what you want. Head to Skills in your workspace to get started.