Creative Thinking

Version: 1.1.0 | Changelog

A structured protocol that shifts an agent's cognitive mode mid-task using semantic temperature control. Three phases: Ground (surface assumptions), Explode (5-level Depth Ladder for escalating creativity), Forge (stress-test and synthesize). Works with any soul.

Price: Free Type: Skill (universal)

What's Included

• SKILL.md - The complete 3-phase creative thinking protocol with Depth Ladder

How It Works

The protocol uses specific language patterns to widen and narrow the token probability distribution at targeted moments. Phase 2's language ("forget constraints," "surprise yourself") functions as a semantic temperature increase. Phase 3's language ("eliminate ruthlessly," "stress test") narrows it back. The agent's soul determines the flavor of creativity. This skill provides the structure.

The Depth Ladder

Five escalating levels:

1. Obvious variations - conventional alternatives (warm-up)
2. Assumption destroyers - violate each assumption from Phase 1
3. Cross-domain theft - steal solutions from unrelated fields
4. The inversion - flip the worst possible approach
5. The absurd leap - the idea you'd normally filter out

Install

1. Copy into your agent's workspace

cp -r creative-thinking/ ~/.openclaw/workspace-{agent-id}/skills/creative-thinking/

Skills must live inside the agent's workspace. Do not copy to a global skills directory.

2. Add to your agent config

In openclaw.json, add the skill to that agent's skills list:

{
  "id": "{agent-id}",
  "skills": [
    "creative-thinking"
  ]
}

3. Use it

Ask your agent to "brainstorm," "think creatively," "explore options," or "think outside the box." The protocol activates automatically.

Recommended Model

Works with any model. Opus produces the widest creative range. Sonnet is solid and cost-effective.