Memory-Induced Tool-Drift in LLM Agents
arXiv:2605.24941v1 Announce Type: cross Abstract: Modern LLM agents combine long-term memory for personalization with tool-calling interfaces for taking actions in the world -- a combination underpinn
Knowledge catalogue
arXiv:2605.24941v1 Announce Type: cross Abstract: Modern LLM agents combine long-term memory for personalization with tool-calling interfaces for taking actions in the world -- a combination underpinn
arXiv:2605.24248v1 Announce Type: cross Abstract: The Model Context Protocol (MCP) standardizes how a large-language-model (LLM) agent and an external tool server exchange messages, but not trust: a h
arXiv:2605.12521v1 Announce Type: cross Abstract: Multi-turn tool calling is essential for LLMs to function as autonomous agents, yet synthesizing the training data required for these capabilities rem
arXiv:2605.13119v1 Announce Type: cross Abstract: Vision-language-action (VLA) models are effective robot action executors, but they remain limited on long-horizon tasks due to the dual burden of exte
arXiv:2510.03992v2 Announce Type: replace-cross Abstract: Large language models (LLMs) are increasingly deployed in agentic systems, where a fundamental task is mapping user intents to relevant extern
arXiv:2604.12896v1 Announce Type: new Abstract: Multimodal language models (MLLMs) are increasingly paired with vision tools (e.g., depth, flow, correspondence) to enhance visual reasoning. However, d
arXiv:2507.03336v4 Announce Type: replace Abstract: Large language models (LLMs) are increasingly tasked with invoking enterprise APIs, yet they routinely falter when near-duplicate tools vie for the
arXiv:2604.11790v1 Announce Type: cross Abstract: Tool-augmented Large Language Model (LLM) agents have demonstrated impressive capabilities in automating complex, multi-step real-world tasks, yet rem
arXiv:2604.07816v1 Announce Type: new Abstract: Tool learning has emerged as a promising paradigm for large language models (LLMs) to address real-world challenges. Due to the extensive and irregularl
arXiv:2604.06185v1 Announce Type: cross Abstract: Fulfilling user needs through Large Language Model multi-turn, multi-step tool-use is rarely a straightforward process. Real user interactions are inh
arXiv:2610.11268v1 Announce Type: new Abstract: Agents combine reasoning with tools to interact with external systems and complete real-world tasks. Early agents typically interleave reasoning and act
arXiv:2510.10931v3 Announce Type: replace Abstract: While reinforcement learning (RL) enhances their ability to plan and reason across retrieval steps, we identify a critical failure mode in this sett
arXiv:2610.06293v1 Announce Type: new Abstract: Multimodal Large Language Models (MLLMs) have demonstrated remarkable potential in video understanding, yet their reliance on retrospective summarizatio
arXiv:2610.01166v1 Announce Type: cross Abstract: Cardiovascular magnetic resonance (CMR), including cine imaging, is a reference standard for the noninvasive assessment of cardiac morphology and vent
arXiv:2610.02206v1 Announce Type: cross Abstract: LLMs are increasingly applied to cybersecurity workflows, where they are expected to translate analysts' intent into tool invocations. However, existi
arXiv:2602.13691v2 Announce Type: replace Abstract: Recent advancements in Large Language Model (LLM) agents have demonstrated strong capabilities in executing complex tasks through tool use. However,
arXiv:2610.01349v1 Announce Type: cross Abstract: Tool-using large language model (LLM) agents turn generated text into real side effects, so poisoned tool metadata, retrieved pages, memory, and reusa
arXiv:2508.08882v5 Announce Type: replace Abstract: Recent advances in multi-agent systems highlight the potential of specialized small agents that collaborate via division of labor. Existing tool-int
A while back I posted here about a CLI agent I'm building that parses tool calls out of a model's plain-text output instead of trusting native function-calling to work. Finally ran the actual numbers.
arXiv:2609.29050v1 Announce Type: new Abstract: Tool-calling agents produce heterogeneous outputs, interleaving structured tool invocations with user-facing natural language summaries. This output het
arXiv:2609.26693v1 Announce Type: new Abstract: A coding agent must emit a valid tool call--a parseable invocation of a tool in the provided schema--before the harness can execute its chosen action. W
arXiv:2512.24565v4 Announce Type: replace Abstract: Large Language Models (LLMs) are increasingly serving as autonomous agents, and their utilization of external tools via the Model Context Protocol (
arXiv:2609.14523v1 Announce Type: new Abstract: Change visual question answering (Change VQA) requires understanding semantic changes across bi-temporal remote sensing images. Although vision language
arXiv:2609.05736v2 Announce Type: new Abstract: LLM tool agents can be improved without retraining by modifying the runtime harness around a fixed model: prompts, tool interfaces, middleware, state ha
arXiv:2609.06815v1 Announce Type: cross Abstract: An open, networked web will allow agents to run frozen models from multiple vendors, keep their history private, and teach each other which tool to ca
arXiv:2608.30952v1 Announce Type: cross Abstract: Chemistry questions often demand exact computation and database lookups that a language model cannot supply from its parameters, so it must reach for
arXiv:2608.25559v1 Announce Type: new Abstract: Video deep research answers complex questions by jointly understanding video content and retrieving external knowledge from the open Web. However, diver
arXiv:2608.20397v1 Announce Type: new Abstract: Agentic large language models (LLMs) on the Model Context Protocol (MCP) re-encode verbose tool schemas every turn, so prefill - quadratic in sequence l
arXiv:2608.19662v1 Announce Type: new Abstract: Agentic language models repeatedly encode tool and skill schemas that recur across requests in different combinations and orders, preventing standard pr
Most Ollama frontends I've tried are chat UIs with tool calling bolted on as an afterthought, which is fine until you actually try to make a small local model do multi-step work and it silently emits
arXiv:2608.10357v1 Announce Type: cross Abstract: Long-horizon tool-using agents must reason over user goals, domain policies, tool calls, simulator state, and delayed verifiable rewards. Reinforcemen
arXiv:2608.06057v1 Announce Type: new Abstract: Tool-calling agents infer task state from accumulated dialogue and tool traces. In persistent interactions, however, historical traces may remain struct
arXiv:2608.03403v1 Announce Type: new Abstract: The performance bottleneck of agents is increasingly shifting from model capability to the robustness of their execution processes. Tools play a central
arXiv:2608.02645v1 Announce Type: cross Abstract: Large Language Model (LLM) agents rely on external tools to perform multistage tasks. Existing agent frameworks typically assume that tool calls are a
If you’re an IT leader, you might be getting a lot of questions about how to build and deploy agents. The pressure to move fast is intense, but the engineering reality is incredibly complex. Where do
arXiv:2607.02873v1 Announce Type: cross Abstract: Large language model agents driving security tool suites over the Model Context Protocol are increasingly common. Yet the factors that bound their cap
arXiv:2604.16870v2 Announce Type: replace-cross Abstract: AI agents increasingly call external tools (file system, network, APIs) through the Model Context Protocol (MCP). These tool calls are the age
arXiv:2606.30632v1 Announce Type: cross Abstract: Can the robot use a plate to cut a cake if no knife is available? Tool use greatly expands robot capabilities, but to use tools creatively beyond thei
arXiv:2606.22251v1 Announce Type: new Abstract: Tactile sensing enables robots to perceive rich contact information at the grasp, supporting tasks such as object recognition, in-hand pose estimation,
arXiv:2606.11702v1 Announce Type: cross Abstract: To make clinically grounded decisions, medical AI agents are expected to go beyond simple recognition and be capable of tool retrieval, evidence acqui
arXiv:2606.10803v1 Announce Type: cross Abstract: Multimodal Large Language Models (MLLMs) excel at utilizing digital APIs and increasingly serve as the 'brain' of embodied AI, instructing robots to i
Uber reportedly now caps coding agents at $1,500/month per employee per tool - seems sensible to me, but it's also an interesting hint at the value Uber thinks these tools are providing https://simonw
arXiv:2606.00566v1 Announce Type: cross Abstract: As language models take on agentic roles that span calling external APIs, reading tool outputs, and acting on instructions embedded in third-party con
Hermes Agent has been updated with a Tool Search feature that enables agents to dynamically identify and load only the tools necessary for a given task, rather than loading all available tools upfront
arXiv:2605.26154v1 Announce Type: cross Abstract: LLM-driven agents are capable of selecting external tools to complete users' tasks. However, attackers could compromise such process, steering agents
arXiv:2605.16909v1 Announce Type: new Abstract: Tool-using agents are increasingly expected to operate across realistic professional workflows, where they must interpret multimodal inputs, coordinate
Together AI launched Voice Finder, a tool designed to help developers quickly select appropriate voices for their applications from a library of over 600 voice options. The tool streamlines the voice
arXiv:2605.06890v1 Announce Type: new Abstract: AI agents are promising for high-stakes enterprise workflows, but dependable deployment remains limited because tool-use failures are difficult to diagn
arXiv:2605.04107v1 Announce Type: cross Abstract: Production agent frameworks (OpenAI Function Calling, Anthropic Tool Use, MCP) transmit tool schemas as JSON, a format designed for machine parsing, n
arXiv:2601.03555v2 Announce Type: replace Abstract: Training reliable tool-augmented agents remains a significant challenge, largely due to the difficulty of credit assignment in multi-step reasoning.
A talk titled 'Every API is a Tool for Agents' from the AI Engineer conference is now available on YouTube, discussing how APIs can be leveraged as tools for AI agents. The presentation was facilitate
Google Cloud Next ‘26 took place this week in Las Vegas, and the energy was incredible as we welcomed over 32,000 leaders, developers, and partners to explore the Agentic Era with us. Across three key
arXiv:2601.20144v3 Announce Type: replace Abstract: Tool-calling agents are increasingly deployed in real-world customer-facing workflows. Yet most studies on tool-calling agents focus on idealized se
arXiv:2604.17886v1 Announce Type: new Abstract: Users often omit essential details in their requests to LLM-based agents, resulting in under-specified inputs for tool use. This poses a fundamental cha
arXiv:2510.22977v2 Announce Type: replace-cross Abstract: Enhancing the reasoning capabilities of Large Language Models (LLMs) is a key strategy for building Agents that 'think then act.' However, rec
arXiv:2609.39066v1 Announce Type: new Abstract: Conventional image forgery detection methods produce binary scores or pixel-level masks without interpretable evidence, while recent multimodal large la
I've liked how Muse-Glimmer worked, so I wanted to see if I could produce a smaller 'kid' out of it. Ornith's sharp decisions on when to think and which tool to call were the other thing I liked, so O
arXiv:2609.31234v1 Announce Type: new Abstract: Problem. Ultra-high-resolution (UHR) remote sensing with vague user intents has two bottlenecks: visual tokens are expensive, and tool calling must be f
arXiv:2609.17985v1 Announce Type: new Abstract: AI agents are usually evaluated by whether they complete a task. In interactive service settings, a successful agent can still frustrate users by asking
arXiv:2609.17848v1 Announce Type: new Abstract: Limited controlled evidence exists on how training data, adaptation method, and model scale jointly affect tool-calling performance in language-model ag