← Resources

AI Innovation Skill

An AI agent skill for structured ideation through the adjacent possible.

Adjacent possible is a structured ideation method: map what exists, find what's one step away, advance along the frontier, then remap. Each iteration reshapes the landscape. The process continues until the frontier thins or a clear idea crystallises.

The skill guides an AI agent through this loop. It works for brainstorming product ideas, exploring architectural possibilities, mapping innovation frontiers in any domain, or finding the next move when you're stuck.

How it works

  1. Map the actual: Lay out what exists right now as concrete elements with properties and relationships. No speculation.
  2. Map the adjacent possible: List what's one step from the actual: composable from existing elements, filling a gap, or recombining what's there.
  3. Advance the frontier: Pick one possibility, explore it concretely, then remap the actual. New elements create new adjacencies.
  4. Iterate or stop: Return to step 2 with the updated landscape. Repeat until the frontier thins or an idea crystallises.

Install

Clone the repository into your OpenCode skills directory:

git clone https://github.com/Organifanten/adjacent-possible.git ~/.config/opencode/skills/adjacent-possible

The skill will be available in your next OpenCode session. Invoke it when you want to brainstorm, explore possibilities, or map an innovation frontier.

Use it

Once installed, the skill activates when you ask OpenCode to explore possibilities or brainstorm. You can also invoke it directly:

What's the adjacent possible for [your topic]?

Source

github.com/Organifanten/adjacent-possible