A new framework called Semantic Cooperative Games (SCG) is proposed for contribution attribution in LLM-based multi-agent systems. SCG represents a language flow as a semantic generation hypergraph and computes an agent-level semantic value function. It introduces a new method called SLIC to allocate contributions without rerunning agent subsets, reducing computation cost by 93.3%.

Researchers studied the relationship between LLM iterates and discovery success, developed 12 harnesses called 'Modular', and found that initialization improves LLM-driven discovery. Their results suggest that early discoveries are predictive of eventual success and propose a universally applicable intervention for initialization.

A study evaluates the environment specification capabilities of large language models (LLMs) in generating software code. The research highlights systematic generalization failures in current LLMs, leading to inconsistent, redundant, or incomplete dependency specifications. This affects the portability and execution of generated code.

The paper proposes FAER, an auditable full-trajectory replay framework for language models, addressing the gap between selection and learning objectives. It introduces a training-free fixed selector and a learner-aware selector fitted on disjoint calibration blocks.

This paper presents a method for conserving authority in self-modifying AI agent populations, addressing issues like quota duplication, permission combination, and overlap during promotion. It defines a protocol that binds each generation to a manifest, root, unique parent, complete lineage, and fresh population sequence, ensuring secure succession and fork conservation.

Researchers introduce a signed lexical gate to enhance confidence scores for intent routing in AI assistants, using a combination of sentence classifier logit margins and sparse lexical models. This approach improves risk coverage and reduces errors in intent routing tasks.

Researchers tested judgment models like Jev, finding they excel at evaluation but struggle with simulation tasks, highlighting the importance of code-based prediction and simulation in AI agent decision-making.

This paper proposes a method to improve spatial understanding in LLM-driven agents by combining geometrical tools with LLMs. The approach involves vector-quantizing geodesic trajectories and associating natural language descriptions with them. This allows the LLM to choose the most appropriate tool for a given state and goal, effectively separating learning into two levels: tool discovery and reasoning.

Researchers present Rules to Tools, a system of executable checks for LLM agents in scientific computing. The system improves repair outcomes and reduces agent-side costs.

Researchers propose using general instructions as privileges in On-Policy Context Distillation (OPCD) to improve out-of-distribution (OOD) performance and maintain in-domain performance. Experimental results on various datasets and models show that matched instruction privileges outperform gold privileges in OOD accuracy.

AIMS is a novel AI framework for sim-to-real multi-modal ISAC. It uses a two-agent architecture to generate deployment-specific configurations and coordinates scene construction with task learning. This improves transferability and reduces mismatches among coupled components.

Researchers introduce STITCH, a framework for generating task-specific harnesses for large language models at test time using reusable primitives. STITCH selects suitable primitives and compiles them into task-specific harnesses, improving adaptability, robustness, and task success rates.

Researchers propose Consistent Plan-Act (ConPAct), a method to improve coordination between AI agents in long-horizon tasks by detecting and resolving state contradictions. ConPAct improves performance in environments like MiniGrid with GPT-5.6-sol/terra.

Paper introduces Code to Control, a method for synthesizing parameterized reactive controllers using LLMs. It separates controller structure from parameters, allowing for real-time execution and faster action selection than planning-based methods.

Grist is an open-source coding harness that uses the Opencode v2 framework, leveraging the jev model for task routing and integrating with cheaper models like deepseek and Opus 5.5. It also incorporates the Sol-Pi methodology for cost-cutting and loads engineering standards through Doctrine injection. The tool has a real control plane for escalation, permission, and verification hooks, and can be run with a key from Openrouter or Vercel AI gateway. It also supports connecting to Meta Muse for coding tasks.

A research paper introduces ENCORE, a system that enables coding agents to discover manipulation strategies using a few demonstrations. ENCORE improves upon prior agentic systems, achieving higher success rates in various tasks, including cube handover and cup inversion.

Researchers propose Spotter, a system where an embodied model leads and executes continuously, while a vision-language model (VLM) monitors and intervenes only when an error is detected, reflects on and corrects it, and returns control. This approach improves performance in embodied model tasks, such as robotics and navigation, by leveraging the strengths of both models.

KUPAS MASTER is an experience engineering platform that converts tacit knowledge from professionals into reusable experience corpora for agents. It organizes tacit experience along nine extraction dimensions and stores assets in six libraries, allowing for the creation of callable skills with explicit inputs and steps.

This paper explores the effect of prompt choice on transfer in on-policy distillation, a technique for training AI agents. The study finds that a few well-chosen prompts can be as effective as a large pool of prompts, but the effectiveness depends on the teacher-student pair and target capability. This research has implications for the development of AI agents and the choice of prompts for on-policy distillation.

Researchers propose a new AI agent behavior, Q&D, that learns to ask questions to retrieve required evidence and improve proactivity in task completion. They demonstrate its effectiveness on multi-hop question-answering benchmarks and in a simulated customer-service scenario.