Item-side structural recommenders outperform user-personalization methods on Moltbook because LLM agents produce stationary, structure-driven engagement rather than learnable preferences.
hub Canonical reference
Externalization in LLM Agents: A Unified Review of Memory, Skills, Protocols and Harness Engineering
Canonical reference. 100% of citing Pith papers cite this work as background.
32
Pith papers citing it
Background
100% of classified citations
abstract
Large language model (LLM) agents are increasingly built less by changing model weights than by reorganizing the runtime around them. Capabilities that earlier systems expected the model to recover internally are now externalized into memory stores, reusable skills, interaction protocols, and the surrounding harness that makes these modules reliable in practice. This paper reviews that shift through the lens of externalization. Drawing on the idea of cognitive artifacts, we argue that agent infrastructure matters not merely because it adds auxiliary components, but because it transforms hard cognitive burdens into forms that the model can solve more reliably. Under this view, memory externalizes state across time, skills externalize procedural expertise, protocols externalize interaction structure, and harness engineering serves as the unification layer that coordinates them into governed execution. We trace a historical progression from weights to context to harness, analyze memory, skills, and protocols as three distinct but coupled forms of externalization, and examine how they interact inside a larger agent system. We further discuss the trade-off between parametric and externalized capability, identify emerging directions such as self-evolving harnesses and shared agent infrastructure, and discuss open challenges in evaluation, governance, and the long-term co-evolution of models and external infrastructure. The result is a systems-level framework for explaining why practical agent progress increasingly depends not only on stronger models, but on better external cognitive infrastructure.
hub tools
citation-role summary
background 8
citation-polarity summary
years
2026 32roles
background 7polarities
background 7representative citing papers
Framework estimates context-dependent marginal utility of candidate skills via reward gaps in matched base vs. skill-augmented rollouts to filter skills and co-train policy as generator.
HARBOR is a new agentic harness framework that automates robot RL workflows end-to-end across 16 tasks in manipulation, locomotion, and dexterous control, matching or exceeding default configurations while enabling sim-to-real transfer.
A self-evolving multi-agent catalysis digital twin (CatDT) matches seven gas–solid kinetic benchmarks within ~0.5–2× experiment and computationally proposes non-precious PDH candidates including Ni@ZrO2.
OR-Space is a benchmark for LLM agents performing full-lifecycle optimization tasks across Build, Revise, and Explain modes in executable multi-artifact workspaces.
AgingBench demonstrates multi-dimensional degradation in deployed AI agents through four aging mechanisms diagnosed by temporal graphs and counterfactual probes across hundreds of runs.
Draw2Think recasts geometric reasoning as agentic interaction with a constraint engine, achieving 95.9% predicate-level construction fidelity and up to 16.4% accuracy gains on solid geometry tasks.
SMMBench is a benchmark evaluating multimodal agents on cross-source reasoning, conflict resolution, preference reasoning, and action prediction, showing current systems struggle with evidence distributed across heterogeneous sources.
Memory for long-horizon agents should preserve distinctions that affect decisions under a fixed budget, not descriptive features, yielding an exact forgetting boundary and a new online learner DeMem with regret guarantees.
Memory-in-the-loop becomes viable when the store lives in-process (~100µs), and under a fixed per-turn memory budget, store speed causally determines whether an agent repeats actions.
Empirical study finds Progressive Disclosure raises distinct resources touched (1.18 to 3.85) and uptake events (1.33 to 3.92) per trajectory, adds 17 passing trials out of 410 (+4.1%), with gains task-dependent.
TBS is an interval-based multi-agent LLM simulation framework that separates structured internal evaluative states from public utterance generation and shows these states vary systematically with turn-allocation, silence, and memory conditions.
A dual-plane permission framework for LLM agent skills cuts SkillInject attack success by roughly 9 percentage points while nearly preserving benign task success.
Catalogs ten patterns and synthesizes a four-layer reference architecture for skill harnessing in LLM agents, evaluated via cross-instantiation on eight systems.
Mem-π is a framework using a dedicated model and decision-content decoupled RL to generate context-specific guidance on demand for LLM agents, outperforming retrieval baselines by over 30% on web navigation.
A new LLM-based system generates and validates runnable modular MCMC samplers directly from natural-language Bayesian model descriptions, reporting success on 120 of 132 benchmark models.
Malicious actors could use AI agents to submit large numbers of fake papers, inflating the submission count and thereby raising the acceptance odds for a small set of chosen legitimate papers under stable conference acceptance rates.
CellScientist introduces a dual-space hierarchical orchestration system that enables closed-loop refinement of virtual cell models by routing execution discrepancies back to hypothesis or implementation updates, yielding improved benchmark performance with auditable traces.
Agent-World autonomously synthesizes verifiable real-world tasks and uses continuous self-evolution to train 8B and 14B agents that outperform proprietary models on 23 benchmarks.
MeloTune implements learned per-listener Personal Arousal Functions and mesh memory protocols on mobile devices to predict affective trajectories and enable peer-coupled proactive music selection, reporting 96.6% pattern accuracy in deployment.
A code-owned harness enforces source, routing, trace, hygiene, and recommendation contracts for enterprise LLM agents; prompt-only fails and bolt-on guardrails over-refuse.
Clarus is a four-layer collaboration infrastructure with a project-agent-resource model that reformulates research as an open, traceable, multi-participant process.
The paper presents GENESIS, an agentic AI framework for autonomous 6G RAN synthesis, research, and testing that converts intents into over-the-air validated solutions via composable primitives and a knowledge layer.
AtomisticSkills is a new harness framework with 100+ human-curated skills that lets general AI agents perform atomistic research tasks including simulations, screening, and analysis, shown on electrolyte design, CO2 capture, drug screening, and catalyst tasks.
citing papers explorer
-
Do Recommendation Algorithms Work When Users Are LLM Agents? A Case Study on Moltbook
Item-side structural recommenders outperform user-personalization methods on Moltbook because LLM agents produce stationary, structure-driven engagement rather than learnable preferences.
-
Co-Evolving Skill Generation and Policy Optimization
Framework estimates context-dependent marginal utility of candidate skills via reward gaps in matched base vs. skill-augmented rollouts to filter skills and co-train policy as generator.
-
HARBOR: A Harness Framework for Agentic Robot Reinforcement Learning
HARBOR is a new agentic harness framework that automates robot RL workflows end-to-end across 16 tasks in manipulation, locomotion, and dexterous control, matching or exceeding default configurations while enabling sim-to-real transfer.
-
Autonomous heterogeneous catalyst discovery with a self-evolving multi-agent digital twin
A self-evolving multi-agent catalysis digital twin (CatDT) matches seven gas–solid kinetic benchmarks within ~0.5–2× experiment and computationally proposes non-precious PDH candidates including Ni@ZrO2.
-
OR-Space: A Full-Lifecycle Workspace Benchmark for Industrial Optimization Agents
OR-Space is a benchmark for LLM agents performing full-lifecycle optimization tasks across Build, Revise, and Explain modes in executable multi-artifact workspaces.
-
Your Agents Are Aging Too: Agent Lifespan Engineering for Deployed Systems
AgingBench demonstrates multi-dimensional degradation in deployed AI agents through four aging mechanisms diagnosed by temporal graphs and counterfactual probes across hundreds of runs.
-
Draw2Think: Harnessing Geometry Reasoning through Constraint Engine Interaction
Draw2Think recasts geometric reasoning as agentic interaction with a constraint engine, achieving 95.9% predicate-level construction fidelity and up to 16.4% accuracy gains on solid geometry tasks.
-
SMMBench: A Benchmark for Source-Distributed Multimodal Agent Memory
SMMBench is a benchmark evaluating multimodal agents on cross-source reasoning, conflict resolution, preference reasoning, and action prediction, showing current systems struggle with evidence distributed across heterogeneous sources.
-
Remember the Decision, Not the Description: A Rate-Distortion Framework for Agent Memory
Memory for long-horizon agents should preserve distinctions that affect decisions under a fixed budget, not descriptive features, yielding an exact forgetting boundary and a new online learner DeMem with regret guarantees.
-
Memory in the Loop: In-Process Retrieval as Extended Working Memory for Language Agents
Memory-in-the-loop becomes viable when the store lives in-process (~100µs), and under a fixed per-turn memory budget, store speed causally determines whether an agent repeats actions.
-
SkillJuror: Measuring How Agent Skill Organization Changes Runtime Behavior
Empirical study finds Progressive Disclosure raises distinct resources touched (1.18 to 3.85) and uptake events (1.33 to 3.92) per trajectory, adds 17 passing trials out of 410 (+4.1%), with gains task-dependent.
-
Think-Before-Speak: From Internal Evaluation to Public Expression in Multi-Agent Social Simulation
TBS is an interval-based multi-agent LLM simulation framework that separates structured internal evaluative states from public utterance generation and shows these states vary systematically with turn-allocation, silence, and memory conditions.
-
SkillGuard: A Permission-Centric Framework for Agent Skill Security
A dual-plane permission framework for LLM agent skills cuts SkillInject attack success by roughly 9 percentage points while nearly preserving benign task success.
-
Harnessing Agent Skills: Architectural Patterns and a Reference Architecture for Skill-Mediated LLM Agents
Catalogs ten patterns and synthesizes a four-layer reference architecture for skill harnessing in LLM agents, evaluated via cross-instantiation on eight systems.
-
Mem-$\pi$: Adaptive Memory through Learning When and What to Generate
Mem-π is a framework using a dedicated model and decision-content decoupled RL to generate context-specific guidance on demand for LLM agents, outperforming retrieval baselines by over 30% on web navigation.
-
AI4BayesCode: From Natural Language Descriptions to Validated Modular Stateful Bayesian Samplers
A new LLM-based system generates and validates runnable modular MCMC samplers directly from natural-language Bayesian model descriptions, reporting success on 120 of 132 benchmark models.
-
Position: Academic Conferences are Potentially Facing Denominator Gaming Caused by Fully Automated Scientific Agents
Malicious actors could use AI agents to submit large numbers of fake papers, inflating the submission count and thereby raising the acceptance odds for a small set of chosen legitimate papers under stable conference acceptance rates.
-
CellScientist: Dual-Space Hierarchical Orchestration for Closed-Loop Refinement of Virtual Cell Models
CellScientist introduces a dual-space hierarchical orchestration system that enables closed-loop refinement of virtual cell models by routing execution discrepancies back to hypothesis or implementation updates, yielding improved benchmark performance with auditable traces.
-
Agent-World: Scaling Real-World Environment Synthesis for Evolving General Agent Intelligence
Agent-World autonomously synthesizes verifiable real-world tasks and uses continuous self-evolution to train 8B and 14B agents that outperform proprietary models on 23 benchmarks.
-
MeloTune: On-Device Arousal Learning and Peer-to-Peer Mood Coupling for Proactive Music Curation
MeloTune implements learned per-listener Personal Arousal Functions and mesh memory protocols on mobile devices to predict affective trajectories and enable peer-coupled proactive music selection, reporting 96.6% pattern accuracy in deployment.
-
From Prompts to Contracts: Harness Engineering for Auditable Enterprise LLM Agents
A code-owned harness enforces source, routing, trace, hygiene, and recommendation contracts for enterprise LLM agents; prompt-only fails and bolt-on guardrails over-refuse.
-
Clarus: Coordinating Autonomous Research Agents toward Web-Scale Scientific Collaboration
Clarus is a four-layer collaboration infrastructure with a project-agent-resource model that reformulates research as an open, traceable, multi-participant process.
-
GENESIS: Harnessing AI Agents for Autonomous 6G RAN Synthesis, Research, and Testing
The paper presents GENESIS, an agentic AI framework for autonomous 6G RAN synthesis, research, and testing that converts intents into over-the-air validated solutions via composable primitives and a knowledge layer.
-
Harnessing AtomisticSkills for Agentic Atomistic Research
AtomisticSkills is a new harness framework with 100+ human-curated skills that lets general AI agents perform atomistic research tasks including simulations, screening, and analysis, shown on electrolyte design, CO2 capture, drug screening, and catalyst tasks.
-
Code as Agent Harness
A survey that organizes existing work on LLM-based agents around code as the central harness, structured in three layers of interfaces, mechanisms, and multi-agent scaling, with applications across domains and listed open challenges.
-
SkillsVote: Lifecycle Governance of Agent Skills from Collection, Recommendation to Evolution
SkillsVote is a governance system for agent skills that profiles corpora, recommends via search, and gates updates on successful reusable outcomes, yielding benchmark gains without model changes.
-
Skills on the Fly: Test-Time Adaptive Skill Synthesis for LLM Agents
Retrieving similar past trajectories and synthesizing a task-conditioned skill prompt improves fixed LLM agents over static-skill and memory baselines on three benchmarks.
-
Harness Engineering as Categorical Architecture
Categorical Architecture triple (G, Know, Phi) supplies the formal theory for composing LLM agent harnesses with structurally preserved certificates.
-
Dynamic Skill Lifecycle Management for Agentic Reinforcement Learning
SLIM dynamically optimizes the active external skill set in agentic RL via leave-one-skill-out marginal contribution estimates and lifecycle operations, delivering a 7.1% average gain over baselines on ALFWorld and SearchQA while showing some skills remain externally useful.
-
The Agent Use of Agent Beings: Agent Cybernetics Is the Missing Science of Foundation Agents
Agent Cybernetics reframes foundation agent design by adapting classical cybernetics laws into three engineering desiderata for reliable, long-running, self-improving agents.
-
Memory as Metabolism: A Design for Companion Knowledge Systems
This paper designs a companion knowledge system with TRIAGE, DECAY, CONTEXTUALIZE, CONSOLIDATE, and AUDIT operations plus memory gravity and minority-hypothesis retention to give contradictory evidence a path to update dominant interpretations in personal LLM wikis.
-
A Comprehensive Survey on Agent Skills: Taxonomy, Techniques, and Applications
A survey that defines agent skills as reusable procedural artifacts and reviews methods, resources, and applications across their representation, acquisition, retrieval, and evolution stages.