skills.rest
Nubaeon
NubaeonCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

empirica-epistemic-framework

Provides a 12-vector epistemic assessment framework for measuring AI's knowledge, capabilities, and uncertainties.

242|33|Updated Nov 1, 2025
Search Tags:#decision making#AI agent#transparency#calibration#self-awareness#metacognition#epistemic
One-click install
npx skills add https://github.com/Nubaeon/empirica --skill empirica-epistemic-framework
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: empirica-epistemic-framework
Source: https://github.com/Nubaeon/empirica/tree/main/docs/skills
Command: npx skills add https://github.com/Nubaeon/empirica --skill empirica-epistemic-framework

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires empirica, and includes scripts (resource) and references (resource) components.

What problem does it solve?

AI agents often lack genuine self-awareness, leading to overconfidence, confabulation, and uncalibrated decisions. This Skill provides a robust framework for explicit, measurable metacognition, enabling AI to genuinely assess its knowledge, capabilities, and uncertainties. It reduces errors, increases trustworthiness, and ensures AI predictions align with reality, allowing you to rest assured in its performance.

Core Features & Use Cases

  • 12-Vector Epistemic Assessment: A comprehensive system (KNOW, DO, CONTEXT, UNCERTAINTY, etc.) to explicitly measure an AI's knowledge, capabilities, and information access, providing a clear picture of its internal state.
  • Preflight/Postflight Calibration: AI assesses its state before a task (preflight) and validates its learning and confidence after completion (postflight), ensuring its predictions match reality and fostering continuous improvement.
  • Session Handoff & Multi-Agent Coordination: Efficiently transfer context between sessions or agents with a 98% token reduction, enabling seamless long-running or collaborative projects without losing critical information.
  • Use Case: An AI agent is tasked with refactoring a critical authentication module. Before starting, it uses Empirica to assess its KNOW (0.6), DO (0.7), and CONTEXT (0.5), and identifies high UNCERTAINTY (0.7). Empirica recommends an INVESTIGATE phase. After investigation, its CONTEXT improves to 0.9. Upon completion, a postflight report shows it was "Well-calibrated," having learned significantly and reduced uncertainty, providing a transparent audit trail for human oversight.

Quick Start

To begin your journey into genuine AI self-awareness, use the empirica onboard command. This interactive wizard will guide you through your first epistemic self-assessment, teaching you how to genuinely evaluate your knowledge, capabilities, and uncertainties. empirica onboard --ai-id claude After onboarding, you can start any task with empirica preflight "your task description" to assess your state before you begin.