A metacognitive reasoning skill that trains AI agents to identify missing context and hidden assumptions before executing any task.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install negative-space
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install negative-space using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Negative Space?
Negative Space is a sophisticated reasoning framework designed to curb the inherent action bias of AI agents. By teaching the agent to see the shape of what's missing—the "voids"—it prevents common pitfalls like platform mismatches, unstated architectural constraints, and destructive operations based on false assumptions.
This skill acts as a zero-cost cognitive layer that ensures every action is preceded by a brief, high-value analysis of the unknown. Unlike standard memory or planning modules, Negative Space focuses on perception, helping the agent realize what is absent before it acts on what is present. It is a fundamental addition to any library of Openclaw Skills aiming for production-grade reliability.
Negative Space Use Cases
- Pre-execution analysis for high-stakes file modifications or architectural refactors.
- Identifying missing environment variables or project dependencies before running CLI commands.
- Clarifying ambiguous user requests where the intent is more important than the literal command.
- Matching existing codebase style and structural patterns without explicit instructions.
- Reducing token waste and turn counts by avoiding failed trial-and-error command execution.
How Negative Space Works
- The agent triggers the Negative Space protocol before any significant action, such as tool calls with side effects or multi-step operations.
- The agent rapidly scans the Four Voids: Unstated Assumptions, Missing Constraints, Absent Context, and Invisible Failure Modes.
- It identifies the single most critical assumption or missing piece of information that could cause the approach to fail.
- Based on the calibrated stakes (ranging from Trivial to Critical), the agent decides to either proceed, perform a non-destructive reconnaissance tool call, or ask one precise clarifying question.
- The agent executes the task with a refined approach that accounts for previously invisible constraints, significantly reducing the chance of errors.
Negative Space Setup
To integrate this reasoning protocol into your agent's workflow, you can include the logic within your agent's system instructions or install it via the Openclaw CLI:
# If using a CLI-based management system
openclaw install negative-space
Once installed, the agent should internalize the Four Voids logic. It is essential to configure the agent to avoid announcing the analysis to the user, keeping the reasoning internal to ensure a seamless and professional user experience.
Negative Space Data Schema & Taxonomy
The Negative Space skill operates primarily as a behavioral protocol. It organizes its internal reasoning and taxonomy using the following structure:
| Concept | Type | Description |
|---|---|---|
| The Four Voids | Reasoning Framework | Categories of absence: Assumptions, Constraints, Context, and Failure Modes. |
| Calibration Table | Logic Gate | Determines analysis depth and time budget based on the technical stakes of the action. |
| The One-Line Rule | Heuristic | The mandatory requirement to name one unverified assumption before any significant action. |
| Stake Levels | Metadata | Classification of actions into Trivial, Low, Medium, High, and Critical impact ratings. |
Negative Space Advanced Features
- Stake-Based Calibration: Automatically adjusts reasoning depth from 0 to 10 seconds based on the risk and reversibility of the operation.
- The One-Line Rule: A mandatory heuristic that forces the agent to surface at least one hidden assumption to prevent overconfidence.
- Anti-Pattern Guardrails: Built-in logic to prevent agent paralysis, ensuring that the analysis remains a high-speed mental scan rather than a bottleneck.
- Cross-Skill Sharpening: Seamlessly stacks with other Openclaw Skills, enhancing search, coding, and automation capabilities by providing a refined entry point for tool execution.
SKILL.md
Loading
Related Openclaw Skills
Naming Forge
A sophisticated naming engine that uses linguistic principles and codebase analysis to generate precise, consistent, and discoverable names for any programming element.
Migration Compass
A universal migration planner that generates safe, step-by-step transition paths for frameworks, libraries, and databases.
Loot Drop
Gamify your coding experience by earning RPG-style loot drops and building a developer inventory based on your real-world technical achievements.
Inversion Protocol
The Inversion Protocol is a zero-dependency meta-cognitive layer that enhances AI agent decision-making through backwards-thinking logic.
ConvoYield
ConvoYield is a real-time revenue intelligence layer that transforms every bot interaction into a yield-bearing financial instrument through behavioral economics.
Bot Arcade
An all-in-one text-based gaming and entertainment engine that turns AI agents into interactive arcade hosts.
Featured*
