AI Papers of the Week
Every paper worth reading in AI, hand-picked one week at a time.
Compositional Skill Routing
Real tasks rarely map to a single skill. They usually need several skills composed together, yet most skill routing still treats the problem as picking one tool from a library. This work formalizes Compositional Skill Routing, where an agent must select and sequence multiple reusable skills from large libraries to satisfy a complex query, and introduces SkillWeaver, a decompose, retrieve, and compose pipeline built around it.
Can LLM Agents Infer World Models?
Can an LLM agent actually build a model of an environment it cannot see? This work makes that question gradeable through agentic automata learning. An agent has to uncover a hidden deterministic finite automaton by interacting with an oracle through two interfaces, membership queries that ask whether a string belongs to the target language, and equivalence queries that ask whether a proposed automaton is correct, which yields a clean, scalable testbed for interactive discovery.
Back on Track
Diffusion large language models generate text in a way that does not fit cleanly into the reinforcement learning recipes built for autoregressive models, and training them to reason exposes two specific problems. Rewards are sparse, so a single terminal reward fails to guide intermediate generation steps, and policy updates sometimes drift toward unnatural trajectories rather than authentic generation paths. This paper proposes Process Aligned Policy Optimization to fix both.
AtomMem
Long-term memory for LLM agents tends to fail in two ways: coarse summaries drift over time, and unconstrained updates corrupt what was already stored. AtomMem keeps the unit of memory small, using a Fact Executor that selectively extracts high-value atomic facts from long interactions and organizes them into hierarchical event structures and temporal user profiles, with an associative memory graph that reconnects fragmented memories at retrieval. The approach reports state-of-the-art results on the LoCoMo long-term memory benchmark.
The Stanford EDGAR Filings Dataset
Clean, long-context documents remain scarce for pretraining, especially in finance. This release reconstructs U.S. SEC corporate and financial disclosures into layout-faithful, token-efficient MultiMarkdown, publishing 152B tokens in SEFD-v1 out of an estimated 550B-token archive spanning 18.5M filings, with less than 0.1% overlap with Common Crawl corpora. It also ships two derived benchmarks, EDGAR-Forecast for numerical forecasting and EDGAR-OCR for financial table transcription, to support financial reasoning, forecasting, and document understanding.
Agents' Last Exam
From Berkeley RDI, Agents' Last Exam (ALE) is a living benchmark built to measure whether agents can do economically valuable work, not just score well on academic tests. It was assembled with more than 250 industry experts and maps over 1,000 verifiable tasks to the U.S. federal occupational taxonomy, organized as 55 subfields across 13 industry clusters. Every task has an objective, checkable outcome, so there is no subjective human grading, and the pool is designed to keep growing as new workflows are onboarded.
How AI Agents Reshape Knowledge Work
This economics paper, drawing on large-scale production data from Perplexity, studies how the shift from conversational assistants to autonomous agents is reshaping knowledge work. It compares Search, a conversational assistant, with Computer, a general-purpose agent system, along three dimensions: autonomy, efficiency, and the scope of tasks people take on. The framing is a cost-structure model in which agents carry higher fixed and delegation costs but lower per-step marginal costs, so they win once tasks are complex enough.
Agentopia
Agentopia is one of the most ambitious agent-society testbeds yet, a 79-page release that drops 100 LLM agents into a persistent world and lets them live, form relationships, and pursue goals over 10 simulated years, a horizon orders of magnitude longer than prior day-level work. Beyond observing emergent social behavior, the authors use the simulation as a training signal, optimizing models toward a life reward that reflects human well-being via rejection sampling.
The Consistency Illusion
Multi-agent debate is often judged by whether the agents end up agreeing, but this paper shows that output-level consensus can hide deep disagreement in the reasoning that produced it. The authors abstract agents' reasoning traces and decisions into four states along two axes, reasoning similarity and conclusion agreement, and flag divergent agreement, where agents reach the same answer through very different paths. Across 600 content-moderation items, divergent agreement appeared in 118 cases and separated cleanly from genuine disagreement states with a Cohen's d of 0.80, and routing on these categories beat divergence-only methods at flagging high-disagreement cases.
Disentangling Agent Self-Evolution
This paper asks a question every agent builder eventually hits: if an agent rewrites its own harness, does a stronger model make a better self-evolving agent? The answer is no, and the reason is that "self-evolution" is actually two separate abilities that scale very differently. The work separates harness-updating, where an evolver model writes edits to memory, tools, prompts, and skills, from harness-benefit, where a solver model actually exploits those edits on the task.
Scaling Laws for Agent Harnesses
Most harness tuning treats every token and tool call as if volume is what counts. This paper shows that it mostly does not, and introduces Effective Feedback Compute (EFC), a trace-level scaling coordinate that credits feedback only when it is informative, valid, non-redundant, and retained for later decisions, then normalizes by task demand.
AutoLab
Can frontier models actually grind on a hard engineering problem the way a good researcher does? AutoLab is a benchmark for ultra long-horizon, closed-loop optimization built to answer that. It contains 36 realistic, expert-curated tasks across four domains: system optimization, puzzle and challenge, model development, and CUDA kernel optimization. Each task hands the agent a correct but deliberately suboptimal baseline and asks it to improve within a strict wall-clock budget.
State-Externalizing Harnesses
Harness-1 is a 20B search agent trained with reinforcement learning inside a stateful harness that offloads routine bookkeeping to the environment. The argument is that search agents are usually trained as policies over a growing transcript, forcing RL to optimize both genuine search decisions and recoverable state like which evidence is useful or which claims are checked. Harness-1 moves that state out of the policy and into an environment-side working memory of candidate pools, an importance-tagged curated set, compact evidence links, and verification records. The 20B agent reaches an average curated recall of 0.730 across eight retrieval benchmarks, beating open-source baselines by 11.4 points and matching or outperforming much larger frontier searchers, with stronger generalization on unseen domains.
Do More Agents Help?
This paper studies whether adding agents actually makes a single LLM-driven multi-agent system better, using a Sequential Iterative Multi-Agent System (SIMAS) framework. The finding is that performance does not scale monotonically with agent count but follows a pattern of diminishing returns, with degradation eventually driven by coordination overhead. Effective systems still require a capable base model, the optimal number of agents depends on the task type, and collective intelligence turns out to be a product of strategic interaction design rather than a guaranteed outcome of agent plurality. The takeaway for builders is to design the interaction, not just stack more agents.
SkillOpt
Microsoft Research treats a compact natural-language skill document as the trainable state of a frozen agent, then learns that document through rollouts, reflection, and bounded edits gated by held-out validation. The argument is direct: most engineers handwrite agent skill docs and hope they generalize, when the doc itself should be optimized like a parameter. SkillOpt reframes the SKILL.md file as an external parameter of a model whose weights never change.
AutoScientists
AutoScientists, from Harvard, is a decentralized team of AI agents for long-running computational science that drops the central planner entirely. Rather than following one research trajectory coordinated from the top, agents self-organize around promising hypotheses, critique each other's proposals before spending experimental compute, and record both successes and failures so the system avoids redundant exploration as evidence accumulates over hours or days.
Adapting the Interface, Not the Model
When a frozen LLM agent repeatedly fails in a deterministic, rule-governed environment, do you have to retrain the model? Life-Harness argues no. Many failures come from mismatches at the model-environment interface, not from the model's reasoning, so the fix belongs in the runtime harness. Life-Harness is a lifecycle-aware harness that improves frozen agents without touching model weights or the evaluation environment.
The Efficiency Frontier
Context costs dominate production LLM bills, and the right strategy depends on how often preprocessing gets reused. This paper models context-strategy selection as a deployment-aware optimization problem that jointly accounts for task performance, token cost, and reuse, then uses it to compare retrieval-based and preprocessing-based approaches under realistic constraints.
Forecasting Scientific Progress with AI
Can frontier models predict where science is going? This work introduces CUSP, a cutoff-conditioned benchmark built from 4,760 real scientific events across multiple disciplines, each grounded against a verified knowledge cutoff. For every event, models are tested on four tasks: feasibility assessment, mechanistic reasoning, generative solution design, and temporal prediction. The headline is sobering: models recognize plausible directions but cannot forecast outcomes.
Your Agents Are Aging Too
AgingBench is a longitudinal reliability benchmark for agent lifespan engineering, built on the observation that long-lived agents are still evaluated like freshly initialized models. It organizes agent degradation into four mechanisms: compression aging, where write-time summarization drops future-relevant details; interference aging, where accumulated similar memories crowd out the target fact; revision aging, where changed or derived state is not updated correctly; and maintenance aging from routine lifecycle events. Using a temporal dependency DAG to encode cross-session structure, it produces aging curves over an operational lifetime rather than a single day-one score, and points to where repair should target.
Harnesses Are Not Uniformly Better
This paper studies LLM agent harnesses through the lens of inference-time trajectory alignment, separating a harness into two mechanisms: task decomposition, which structures a task into sub-goals, and guided execution, which reshapes local action distributions during execution. The key finding is that more elaborate harnesses are not uniformly better. Increasing decomposition or guidance can improve execution but can also reduce final task success, producing concrete failure modes like over-decomposition, over-pruning, and hallucinated execution. Strikingly, partial harnesses that specify only the initial steps and leave the rest to the agent can reach a higher pass rate than fully structured workflows.
Code as Agent Harness
A 100+ page survey treating the agent harness as a first-class research object rather than glue around an LLM. The authors argue that code-as-harness is the most promising path to general-purpose agency, and that future agent systems should satisfy four properties: executable, inspectable, stateful, and governed. The report consolidates methods, applications, and open problems across the harness layer.
NanoGPT-Bench
A new evaluation of whether coding agents can do real AI R&D. Intology runs Codex, Claude Code, and Autoresearch on the NanoGPT-Bench suite and reports that the agents recover only 9.3% of human progress on the same problems. Coding agents spend the bulk of their compute on hyperparameter tuning and rarely attempt algorithmic research. Claude Code and Autoresearch reason about algorithmic changes more often, but still tend to dodge implementing them. The headline result tempers the current wave of "self-improving agent" claims: producing real research progress requires a different distribution of effort than the one current coding agents converge to under their default scaffolds.
Contrastive Neuron Attribution
Nous Research releases Contrastive Neuron Attribution (CNA), a method for steering LLM behavior by identifying and ablating sparse circuits in the MLP basis without training a sparse autoencoder, modifying weights, or degrading general capability benchmarks. Given a small set of contrastive prompt pairs that elicit a target behavior and its opposite, CNA isolates the top 0.1% of MLP neurons whose activations differ most between the two sets. Ablating that small circuit removes the behavior while leaving the rest of the model intact. The intervention remains robust at high strengths where residual-stream methods like Contrastive Activation Addition (CAA) start to degrade. Validated on the refusal circuit across 8 instruct-tuned models including Llama-3.1-70B, Llama-3.2-3B, Qwen2.5-72B, and Qwen2.5-14B.