A disciplined decision-framing agent skill designed to surface cognitive biases and neutralize emotionally slanted requests before they harden into flawed decisions.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install bias-audit-clarkchenkai
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 bias-audit-clarkchenkai 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 Bias Audit?
Bias Audit is a specialized protocol within the Openclaw Skills ecosystem that provides a rigorous framework for identifying and neutralizing cognitive distortions in decision-making. Rather than providing a generic response to a user's prompt, this skill audits the request for anchoring, framing effects, and confirmation pressure. It is specifically designed to make bias legible without being accusatory, ensuring that AI-driven insights remain grounded in logic rather than emotional momentum.
By implementing this skill, developers and users can ensure their AI agents act as a friction point against 'obvious' conclusions backed by weak evidence. It transforms loaded or binary choices into neutral, evidence-based assessment frames, making it an essential tool for high-stakes pricing, people management, and project strategy decisions.
Bias Audit Use Cases
- Analyzing loaded or emotionally slanted questions to reveal underlying assumptions.
- Breaking down false binary choices that limit creative problem-solving.
- Auditing project decisions driven by recent vivid examples (Availability Heuristic).
- Neutralizing wording that nudges an AI toward a specific, biased answer.
- Standardizing decision-making criteria for team leadership and procurement.
How Bias Audit Works
- Capture Framing: The agent quotes or restates the original request to make existing biases visible to the user.
- Identify Signals: It scans for specific triggers like loss aversion, default-value distortion, and anchoring.
- Neutral Reframe: The agent rewrites the prompt into a cleaner version with fewer hidden assumptions.
- Evidence Discovery: It identifies missing baselines, counter-evidence, or comparisons necessary for an objective view.
- Criteria Definition: The conversation is shifted from emotional momentum into explicit decision criteria.
- Contractual Output: The process concludes with a standardized six-part summary for consistent documentation.
Bias Audit Setup
To add this bias-audit capability to your agent using Openclaw Skills, include the skill definition in your agent's configuration directory:
# Navigate to your agent skills folder
cd your-agent-path/skills/
# Create the bias-audit definition
touch bias-audit.md
Ensure your system prompt references the bias-audit protocol whenever emotionally slanted triggers are detected in the user input.
Bias Audit Data Schema & Taxonomy
The Bias Audit skill adheres to a strict output contract to ensure data consistency across multiple sessions.
| Section | Description |
|---|---|
| Original Framing | A raw capture of the user's initial biased request. |
| Bias Signals | Named distortions identified (e.g., Anchoring, Framing). |
| Neutral Reframe | A sanitized version of the request for the LLM to process. |
| Missing Evidence | A list of data points or perspectives currently absent. |
| Decision Criteria | The logical metrics for making the final choice. |
| Next Step | A concrete, actionable recommendation to move forward. |
Bias Audit Advanced Features
- Strict Output Contract: Guarantees that every audit follows the exact six-part markdown structure for automated parsing.
- Signal Identification: Specifically tuned to detect complex psychological distortions like loss aversion and default-value bias.
- Non-Pathological Auditing: Focuses on framing and technical truth rather than critiquing the user's personality.
- Evidence Gapping: Proactively requests missing baselines to prevent the AI from hallucinating or agreeing with biased evidence.
SKILL.md
Loading
Related Openclaw Skills
Frustration Translator
A sophisticated tool designed to detect emotional charge in user prompts and translate them into clear, actionable technical instructions for AI agents.
Rate Limit Guard
A protective layer for AI agents that prevents 429 rate limit loops and optimizes token usage during API failures.
Podcast Digest
An advanced AI tool that converts podcast audio and transcripts into structured knowledge assets, interactive visual reports, and cross-episode matrices.
Spawner
The Spawner skill enables the delegation of complex, long-running tasks to independent sub-agents using efficient model routing.
MiniMax MMX-CLI
A comprehensive command-line interface for AI agents to access MiniMax's full suite of multimodal capabilities including video, audio, and image generation.
Pilot GitHub Bridge
A bridge that translates GitHub webhook events into actionable Pilot Protocol messages for autonomous agents.
Featured*
