Research | Tandemly

Masking Stale Observations Helps Search Agents -- Until It Doesn't: A Regime Map and Its Mechanism

Zhang, Xu, Li, Zhang, Jiang, Zhang & McAuley · UC San Diego · 2026

Synthesized July 14, 2026 · A systematic sweep over 4B-to-284B models and three retrievers maps when observation masking helps long-horizon search agents and when it collapses performance: an asymmetric inverted-U driven by model-retriever interaction, not either factor alone

Agentic Search · Context Management

Learning to Reason by Analogy via Retrieval-Augmented Reinforcement Fine-Tuning

Xiao, Ma, Chen, Chen, Atreya, Chen & Ordonez · Rice University / Meta · 2026

Synthesized July 10, 2026 · RA-RFT trains a retriever by expected reasoning utility rather than semantic similarity, then fine-tunes via RLVR on the analogous demonstrations. Up to +7.1 points on AIME 2025 over GRPO

Reasoning · Retrieval

A Unified Framework for the Evaluation of LLM Agentic Capabilities

Zhu, Li, Lyu, Luo, Yang, Liu, Hui, Yuan, Sun, Su & Shao · Beijing University of Posts and Telecommunications, Shanghai AI Lab & collaborators · 2026

Synthesized July 9, 2026 · Seven agent benchmarks re-run through one standardized harness and fixed ReAct loop across 15 models, 400K rollouts, and 5B tokens. Scaffold choice and environment volatility alone are sufficient to reorder leaderboards

Agent Evaluation

DeployBench: Benchmarking LLM Agents for Research Artifact Deployment

Wang, Qian, Zhang et al. · Boston University, Northeastern University & University of Texas at Dallas · 2026

Synthesized July 8, 2026 · Four frontier LLM agents tackle 51 research artifact deployment tasks in clean environments. Best pass rate: 51.0%. The dominant failure: agents stopping before verifying the artifact actually ran

Agent Evaluation

Evaluate Agents Without Deploying Them: Off-Policy Evaluation via Diffusion World Models

Liu, Xiong, Zhang & Tang · Emory University & Shanghai Jiao Tong University · 2026

Synthesized July 7, 2026 · A diffusion world model trained on historical interaction logs simulates environment responses for any candidate LLM agent, enabling reliable off-policy performance estimates without live deployment. Solves a structural problem that broke all prior OPE methods for discrete autoregressive agents

Agent Evaluation

Agent Planning Benchmark: A Diagnostic Framework for Planning Capabilities in LLM Agents

Sun, Wang, Song, He, W. Zhang, Y. Liu, Y. Yang & Y. Cheng · Tongji University, Shanghai AI Lab & collaborators · 2026

Synthesized July 6, 2026 · A diagnostic benchmark with 4,209 cases across 22 domains separates planning failure from execution failure and adds infeasibility detection. Across 12 MLLMs, systematic weaknesses in long-horizon planning, tool-noise robustness, and refusal calibration

Agent Evaluation

ContextPilot: Faster Long-Context LLM Inference via Context Reuse

Jiang, Huang, Cheng, Deng, Sun & Mai · University of Edinburgh · 2026

Synthesized July 5, 2026 · A context index intercepts overlapping prompt blocks across requests, reorders them into cache-reusable prefixes, and deduplicates repeated content with quality-preserving annotations. Up to 3x faster prefill, 36% fewer tokens, zero model changes

LLM Inference

Prompt Injection's Impossible Defense: The Contextual Integrity Problem

Abdelnabi & Bagdasarian · University of Massachusetts CICS · 2026

Synthesized June 24, 2026 · An impossibility result shows no defensive norm can block all prompt injections without also blocking flows the legitimate task needs. Data-instruction separation is a partial mitigation, not a fix

Agent Security

Beyond Consensus: When Traces Know More Than Votes

Fadnavis, Kanakaraj & Wyss · Bioscope AI · 2026

Synthesized June 23, 2026 · In mixture-of-agents systems, reading complete reasoning chains from all agents recovers correct answers that majority voting discards, even when every agent agreed on the wrong answer

Multi-Agent Systems

VibeSearchBench: When Real Search Meets a Benchmark

Xiaohongshu Inc. · 2026

Synthesized June 22, 2026 · A 200-task bilingual benchmark for proactive, multi-turn search agents. The best frontier model scores 30.3 Triplet F1, revealing how far current search agents fall short when handling real-world intent ambiguity

Agent Evaluation

MUSE-Autoskill: The Skill Lifecycle Agents Were Missing

Lin, Li, Song, Jiang & Zhang · ByteDance · 2026

Synthesized June 21, 2026 · A five-component lifecycle for LLM agent skill libraries: create skills from successful tasks, gate admission with unit tests, refine from new examples, retire underperformers. Auto-generated skills hit 68.40% overall and beat the human-skill ceiling on tasks where generation succeeded

Agent Architecture

Is Agent Memory a Database? The Missing Data-Foundations Layer

Orogat & Mansour · Concordia University · 2026

Synthesized June 20, 2026 · Long-term agent memory is a data-management workload, not just a retrieval problem. Existing vector stores support ingest and retrieve but lack a revise operator, and that gap explains most agent memory failures. GEM formalizes four state-level operators replacing CRUD

Agent Memory

Compute Where it Counts: Per-Token Efficiency in Frozen LLMs

Akhauri & Abdelfattah · Cornell University · 2026

Synthesized June 19, 2026 · A lightweight policy reads the LLM's own hidden state at each decode step and picks how sparse, pruned, and quantized that step should be. Up to 7.3-point MMLU gain over uniform compression at matched FLOPs

Inference Optimization

Interpretability Can Be Actionable: A Rubric for the Field

Orgad, Barez, Haklay et al. · multiple institutions · 2026

Synthesized June 15, 2026 · A position paper proposing two criteria for evaluating interpretability work: concreteness of the intervention enabled and validation that the intervention works. Most current work falls short on at least one dimension

Interpretability

Remembering More, Risking More: How Agent Memory Accumulates Safety Risk

Al-Tawaha, Gu, Niu, Jia & Jin · Virginia Tech, UC Berkeley & UIUC · 2026

Synthesized June 14, 2026 · Memory-induced safety violation rates climbed monotonically with exposure length across 8 memory architectures and 3 deployment scenarios. The same agent, same probes, more memory produced more safety violations. Risk is detectable at retrieval time before generation

Agent Safety

δ-mem: Adding Persistent Memory to Any Frozen LLM

Lei, Zhang, Li, Wang et al. · Nanyang Technological University · 2026

Synthesized June 13, 2026 · A compact 8×8 online state matrix uses delta-rule learning to compress past context and inject low-rank corrections into a frozen LLM's attention at each decode step. 1.31× the backbone on MemoryAgentBench, 1.20× on LoCoMo, no base model retraining required

Agent Memory

AgentTrust: A Runtime Safety Layer for Every Tool Call

Yang · Independent · 2026

Synthesized June 8, 2026 · An 8-component runtime pipeline intercepts every agent tool call before execution: deobfuscates shell payloads through nine normalization strategies, tracks multi-step attack chains across sessions via RiskChain, and proposes safer alternatives via 37 SafeFix rules. 95.0% verdict accuracy at low-millisecond latency

Agent Safety

Automated Interpretability: Two Agent Loops, One Autonomous Pipeline

Marin-Llobet & Ferrando · Harvard University · 2026

Synthesized June 7, 2026 · Two coupled agent loops automate mechanistic interpretability end-to-end: a discovery agent navigates activation space via k-NN graphs to surface interpretable features, and an explanation agent iteratively probes and refines hypotheses to beat one-shot auto-interp baselines

Mechanistic Interpretability

AutoTTS: When LLMs Discover Their Own Reasoning Strategies

Zheng, Liu, Huang et al. · UMD, UVA & UNC · 2026

Synthesized June 6, 2026 · A coding agent discovers test-time scaling controllers by searching a frozen reasoning trace cache. The found CMC controller cuts tokens 69.5% vs SC@64 at matched accuracy, for $39.90 discovery cost

Inference Optimization

Dual-Dimensional Consistency: Smarter Self-Consistency Sampling

Xu, Li, Zhao, Wu, Li & Yan · Xi'an Jiaotong University · 2026

Synthesized June 2, 2026 · DDC combines confidence-weighted Bayesian voting and trend-aware path pruning to make self-consistency sampling adaptive. Over 10x token reduction at matched or improved accuracy across five reasoning benchmarks

Inference Optimization

Agentic Systems as Boosting: When Weak Models Beat the Frontier

Sunkaraneni, Beneventano, Neumarker, Poggio & Galanti · MIT & Texas A&M · 2026

Synthesized May 30, 2026 · A committee of nano-scale models hit 76.4% on SWE-bench Verified, matching Gemini 3 Pro and Claude Opus 4.5 Thinking standalone. The framework explains when and why committees of weak models match frontier accuracy: the task needs a local verifier

Agent Architecture

Is Grep All You Need? The Agent Harness Moves Accuracy More Than the Retrieval Method

Sen, Kasturi, Lumer, Gulati, Subbiah et al. · 2026

Synthesized May 28, 2026 · Testing grep vs vector retrieval across four agent harnesses (Chronos, Claude Code, Codex, Gemini CLI) reveals the harness layer shifts accuracy more than the retrieval method; Claude Code favors grep, Gemini CLI favors vector

Agentic Search

BoundaryRouter: Learning When to Escalate to an Agent

Wang, Qiu et al. · Princeton, Michigan, Tsinghua et al. · 2026

Synthesized May 23, 2026 · A training-free cold-start router builds experience memory from a small seed set to route queries between plain LLM inference and full agent execution; 60.6% inference time reduction vs always-agent, 28.6% accuracy gain vs always-LLM

Agent Routing

LaTER: Latent-Phase Reasoning Cuts Tokens 32% Without Losing Accuracy

Li, Wang, Liu et al. · 2026

Synthesized May 18, 2026 · A training-free two-phase method explores in latent space before switching to explicit chain-of-thought. On Qwen3-14B: 32% token reduction, AIME 2025 accuracy improves from 70.0% to 73.3%

Inference Optimization

ComplexMCP: Three Failure Modes in Large-Scale Tool Sandboxes

Li, Yang, Wang et al. · 2026

Synthesized May 17, 2026 · A 150+ tool MCP benchmark exposes three reproducible failure modes in frontier LLM agents: tool-retrieval saturation, over-confidence skipping verification, and strategic defeatism

Agent Evaluation

STALE: When Agent Memory Becomes a Liability

Chao, Bai et al. · 2026

Synthesized May 15, 2026 · A 1,200-query benchmark probes whether frontier LLMs can detect when stored memories have been silently invalidated. Best model scores 55.2%. Implicit conflict is the dominant failure mode

Agent Memory

AI Co-Mathematician: When Scaffolding Beats the Model

Zheng, von Glehn, Zwols et al. · Google DeepMind · 2026

Synthesized May 14, 2026 · The same base model scores 19% alone and 48% inside a multi-agent workbench with parallel workstreams, stored failure records, and enforced review cycles

Agentic AI

Meta-Harness: The 6x Gap Lives in Your Code, Not Your Model

Lee, Nair, Zhang, Lee, Khattab & Finn · Stanford University & MIT · 2026

Synthesized May 13, 2026 · Fixing the model and varying only the surrounding harness code produces a 6x performance gap. Meta-Harness automatically searches for better harness code using full diagnostic history, beating SOTA on text classification, IMO-level math reasoning, and agentic coding

AI Systems

How Coding Agents Actually Perform in the Wild

Popescu, Gros, Botocan, Pandita, Devanbu & Izadi · TU Delft & UC Davis · 2026

Synthesized May 12, 2026 · 110,000 open-source pull requests from five coding agents reveal that agent code gets merged but churns faster than human-authored code over time

Software Engineering

Conversation Reduces Load. Images Build It.

Taneja, Singh & Goel · Georgia Institute of Technology · 2026

Synthesized May 11, 2026 · A 124-person randomized controlled trial found multimodal conversational AI produces better biology learning outcomes than text-only chat or semantic search, explained through two distinct cognitive load mechanisms

AI in Education

Image Generation Diversity: When Models Miss the Map

Dombrowski, Zhang, Cechnicka, Reynaud & Kainz · FAU Erlangen-Nürnberg & Imperial College London · 2025

Synthesized May 10, 2026 · No state-of-the-art image generator covers more than 77% of its training distribution, and standard metrics like FID can't detect the gap. A new metric (IRS) and a model variant (DiADM) address both problems

Generative AI

SLOW: The AI Tutor That Thinks Before It Speaks

Wei, Li & Jiang · Shanghai Institute of AI for Education · 2026

Synthesized May 8, 2026 · A four-stage reasoning workspace that separates cognitive diagnosis, stability validation, affect prediction, and strategy selection before any tutoring response is generated

AI Tutoring

Single-Agent LLMs Outperform Multi-Agent Systems on Multi-Hop Reasoning Under Equal Thinking Token Budgets

Tran & Kiela · Stanford University · 2026

Synthesized May 7, 2026 · When you control for thinking-token compute, the multi-agent advantage on multi-hop reasoning largely disappears across five architectures and three model families

Agent Architecture

LLMs in Games: When Generated Content Runs the Rules

Johnson, Ahmed, Lang, Thethi, Zheng & de Souza Santos · University of Calgary · 2026

Synthesized April 19, 2026 · Student developers embedded LLMs as architectural components in two games. The finding: model errors turned into fairness violations, not cosmetic bugs

Game Development

Arknights: When the AI Lies, Players Learn

Shuai Guo · Uppsala University · 2025

Synthesized April 9, 2026 · A mobile strategy game deliberately gives players unreliable AI guidance, reshaping agency from action to interpretation

Explainable AI

Vibe Coding: Flow, Trust, and Co-Creation

Pimenova, Fakhoury, Bird, Storey & Endres · U Michigan / Microsoft Research · 2025

Synthesized April 4, 2026 · The first qualitative study of vibe coding reveals a new programming paradigm built on flow and calibrated AI trust

Vibe Coding

Games That Teach AI Ethics

Solyst, Nakigozi, Fong & Shapiro · University of Washington · 2025

Synthesized April 3, 2026 · Two multiplayer games use text-to-image AI to teach teens about bias through play

AI Education

BAVT: Spend Less, Reason Better

Li et al. · UBC / Vector Institute · 2026

Synthesized March 25, 2026 · Budget-Aware Value Trees cut AI agent costs by 75% with equal or better accuracy

AI Agents