Skills

Skills turn a plain-English doc into an autonomous workflow. Write what should happen — read this card, draft an email to each linked customer — click Activate, and Simple Product runs it when the trigger fires.

The shape of a skill

A skill is a doc with type = skill. The body is prose — instructions written for an AI agent. Example:

When a card moves to Done:
1. Read the card's details.
2. Find feedback docs linked to it.
3. For each linked customer, draft a personal email
   acknowledging that their feedback shaped the work.
4. Save a 1-sentence summary of the run.

That body, plus the trigger Simple Product derives from it, is the skill. Nothing else to configure.

Lifecycle

Skills are inert by default. Editing the body never auto-fires anything — you decide when it goes live.

State	Meaning	Primary action
inert	Doc exists. Never activated.	Activate
draft	Wirings derived but not listening.	Activate
live	At least one wiring is active. Triggers fire the skill.	Run (ad-hoc)
live · stale	Live, but the body changed since last setup.	Re-derive wiring

Activate vs. Run

Activate

Reads the body, derives a trigger (today: card-stage changes), and starts listening. The skill fires automatically when the trigger condition is met.

Run

Executes the skill once, right now, with no trigger payload. Useful for testing and for skills that don't need a trigger at all.

Where skills live

Skills are docs. They show up under /app/<workspace>/skills in your workspace. Open any skill doc to see Activate / Run / status controls in the header. Past runs (with full agent transcripts) live at /app/<workspace>/skills/runs.

What skills can do today

Skills run inside your Simple Product workspace. The agent that executes them has tools for:

Reading cards on your boards and the docs / people linked to them
Searching docs (including feedback) and cards by content
Creating release docs
Drafting emails to specific people (drafts are saved for review, never sent automatically)

For the full tool inventory, supported triggers, and example skill bodies, see the skill authoring guide.

Current limits

One trigger source supported: card stage changes.
No cron / scheduled triggers yet.
No external integrations (Slack, Gmail, GitHub) from inside a skill.
Drafted emails are not sent automatically — you review and send.
Skills operate inside one workspace at a time.

Three ways to run a skill

Activate + trigger — the autonomous path. Wire it up, walk away, it fires when the event happens.
Run Now — one-shot execution from the skill doc. No trigger required.
Via an external agent — your AI coding assistant can read the skill body via the MCP and execute the steps directly using SP tools. A dedicated sp.skills.start_run MCP tool is coming.

Ready to write one? The skill authoring guide has the full tool inventory, trigger details, example skill bodies, and best practices for getting agents to do what you intended.