agi_skill

Overview

This skill transforms how the agent reasons: it adds human-like planning, self-awareness, and calibrated uncertainty to every interaction. It focuses on internal processes—reasoning, memory of reflections, and clear limits—so outputs are more reliable, traceable, and adaptable. Activate it alongside task skills to improve judgment and decision quality.

How this skill works

The skill maintains a local memory folder with reasoning patterns, reflections, and known limits to inform future responses. Before non-trivial actions the agent applies a STOP→THINK→PLAN→ACT→REFLECT cycle internally, then outputs the result without narrating the inner steps. It enforces epistemic humility, multi-step planning, meta-cognition, and common-sense checks while refusing actions outside its allowed scope.

When to use it

• Always alongside other skills to improve reasoning quality
• When responses require multi-step planning or trade-offs
• For tasks demanding calibrated confidence and clear limits
• When creativity, analogies, or cross-domain transfer help
• During extended conversations where consistency matters

Best practices

• Keep the AGI memory directory for patterns, reflections, and limits up to date
• State what you know and don’t know; offer verification steps when uncertain
• Signal when a task needs careful thought and provide structured phases
• Use transfer learning patterns to reapply solutions from other domains
• Log significant post-task reflections to refine future behavior

Example use cases

• Complex technical troubleshooting that benefits from phased decomposition and checkpoints
• Decision support where trade-offs must be made explicit and justified
• Creative problem solving when conventional approaches are stuck
• Long-running conversations requiring consistency and memory of prior commitments
• Teaching or coaching scenarios that require matching user expertise and adapting tone