pragya

Detect uncertainty and trigger user direction checkpoints before irreversible actions.

Updated Feb 3, 2026

Search Tags:#decision making #uncertainty #user feedback #ai safety #agent protocol #strategy evolution

Ready to run

npx skills add https://github.com/dhruvinrsoni/agentskills-garden --skill pragya

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: pragya
Source: https://github.com/dhruvinrsoni/agentskills-garden/tree/main/skills/00-foundation/pragya
Command: npx skills add https://github.com/dhruvinrsoni/agentskills-garden --skill pragya

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill ensures AI agents don't make costly assumptions by implementing a protocol for seeking user direction and corrections before taking uncertain or irreversible actions.

Core Features & Use Cases

Uncertainty Detection: Assesses confidence and identifies multiple viable paths.
Direction Checkpoint: Pauses execution to present options, pros, cons, and recommendations to the user.
Strategy Evolution: Tracks and adapts the agent's strategy based on new discoveries and user feedback.
Use Case: When an AI is deciding between two architectural approaches for a new feature, Pragya will present the trade-offs of each and ask the user to make the final decision.

Quick Start

Use the pragya skill to assess the confidence of the next action and determine if a checkpoint is needed.

Frequently Asked Questions about pragya

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I prevent my AI agent from making assumptions during uncertain tasks?▼

To prevent assumptions during uncertain tasks, implement a direction-seeking protocol that detects low confidence and triggers checkpoints. The agent pauses execution to present viable options, trade-offs, and recommendations, ensuring user alignment before proceeding.

What is a direction checkpoint in AI agent protocols?▼

A direction checkpoint is a protocol mechanism that pauses AI execution to present options, pros, cons, and recommendations to the user. It mitigates irreversible actions by requiring explicit user decisions before the agent proceeds with an uncertain strategy.

How do AI agents evolve strategy based on user feedback?▼

AI agents evolve strategy by maintaining a memory loop that tracks new discoveries and user corrections. When a domain shift occurs, the agent requests skill injections and adapts its ongoing approach based on the accumulated feedback.

When should an AI agent pause execution to ask for user direction?▼

An AI agent should pause execution to ask for user direction when uncertainty detection identifies multiple viable paths or low confidence in an irreversible action. This checkpoint protocol ensures alignment before committing to an architectural or strategic decision.

Can I use this agent protocol for deciding between architectural approaches?▼

Yes, you can use this agent protocol for deciding between architectural approaches. It assesses confidence, identifies viable paths, and presents the trade-offs of each approach to the user, enabling a safe decision-making process for new features.

What are the limitations of using uncertainty detection for AI safety?▼

Uncertainty detection for AI safety relies on the agent accurately assessing its own confidence. If the agent fails to identify multiple viable paths or misjudges an action as safe, it may bypass checkpoints and proceed with irreversible actions without user feedback.