Write a skill once, use it with any model
Skills are plain markdown and scripts in your folder. Claude and Codex both read them, and so will whatever comes next.
For building with ai
Build on AI without rebuilding every time the tools change.
You already live in Claude Code and Codex. Salana gives that work a home: skills you write once, automations with schedules and triggers, dashboards built by asking, all as plain files you own.
What changes
Automations with a definition of done
A schedule or a trigger, the steps, and what a good run leaves behind. Checked after every run.
Dashboards by asking
Ask for a page and the agent builds it from your files: a board, a chart, a tracker.
Every run leaves a trail
Tasks, outcomes and logs written by the agents themselves, so you can see what happened while you slept.
The memory at work
Everything you do flows into one memory. The agents read it, and the work comes out the other side.
Teach it once. Build on it every time.
Your memory, your routines, any model.