Know When to Trust the Skill: Delayed Appraisal and Epistemic Vigilance for Single-Agent LLMs

TL;DR

This paper introduces MESA-S, a framework for enhancing trust and epistemic vigilance in single-agent LLMs through delayed appraisal and confidence separation.

Contribution

It formalizes a metacognitive architecture that separates self-confidence from source-confidence and incorporates delayed evaluation to improve reliability.

Findings

Explicit trust provenance reduces vulnerabilities.

Delayed escalation prunes unnecessary reasoning.

Decoupling confidence prevents inflation and improves trustworthiness.

Abstract

As large language models (LLMs) transition into autonomous agents integrated with extensive tool ecosystems, traditional routing heuristics increasingly succumb to context pollution and "overthinking". We argue that the bottleneck is not a deficit in algorithmic capability or skill diversity, but the absence of disciplined second-order metacognitive governance. In this paper, our scientific contribution focuses on the computational translation of human cognitive control - specifically, delayed appraisal, epistemic vigilance, and region-of-proximal offloading - into a single-agent architecture. We introduce MESA-S (Metacognitive Skills for Agents, Single-agent), a preliminary framework that shifts scalar confidence estimation into a vector separating self-confidence (parametric certainty) from source-confidence (trust in retrieved external procedures). By formalizing a delayed procedural probe mechanism and introducing Metacognitive Skill Cards, MESA-S decouples the awareness of a skill's utility from its token-intensive execution. Evaluated under an In-Context Static Benchmark Evaluation natively executed via Gemini 3.1 Pro, our early results suggest that explicitly programming trust provenance and delayed escalation mitigates supply-chain vulnerabilities, prunes unnecessary reasoning loops, and prevents offloading-induced confidence inflation. This architecture offers a scientifically cautious, behaviorally anchored step toward reliable, epistemically vigilant single-agent orchestration.