Intuition

What This Skill Does

The Intuition skill enables OpenClaw to bypass standard deliberative reasoning in favor of rapid, pattern-based judgment. Built upon the Recognition-Primed Decision (RPD) model, this skill allows the agent to identify solutions immediately by matching current context against established internal heuristics. When invoked, the agent skips the "analysis paralysis" phase of comparing multiple pros and cons, opting instead to present a single, high-confidence recommendation that mirrors human "gut feelings" gained through extensive domain training.

Installation

To integrate the Intuition skill into your OpenClaw environment, execute the following command in your terminal:

clawhub install openclaw/skills/skills/ivangdavila/intuition

Use Cases

Intuition is highly effective in high-validity environments where patterns are stable and feedback loops are tight. It excels in:

• Code Smell Detection: Instantly identifying sub-optimal logic or security vulnerabilities during code reviews.
• UI/UX Auditing: Making rapid judgments on layout flow, accessibility, and visual hierarchies.
• Content Curation: Determining if a piece of writing matches a specific tone or target audience requirement.
• Communication Dynamics: Recognizing the intent behind complex user messages or team interactions.

Example Prompts

1. "Quick read: Is this SQL query going to bottleneck our production database?"
2. "What's your gut feeling on this UX wireframe—does it feel intuitive for a first-time user?"
3. "First instinct: Which of these two marketing headlines is more likely to convert?"

Tips & Limitations

This skill should be used for speed, not for critical long-term forecasting or rare, novel edge cases where explicit analytical reasoning is mandatory. Always default to the Analytical mode for complex financial or high-risk decision-making. If the agent expresses uncertainty, do not force an intuitive response; instead, trigger a switch to analytical mode. Remember that intuition works best when provided with sufficient historical context. To maintain high accuracy, periodically review the agent's performance in the feedback.md reference file, which tracks accuracy across different domains. Avoid using this skill for tasks that require deep factual verification or complex multi-variable calculations where errors could cause significant downstream failures.