Nobody asks for page two

Three papers about the context an agent actually gets to work with. The first finds that agents handed truncated tool output never go back for the rest, which makes the cut, not the ranking, the whole design decision. The second holds model and tasks fixed, changes only the code deciding what the model sees, and nearly doubles what it solves under a tight window. The third gets a model to abstain on the questions it would have got wrong, without labelling one.

Today's pick

requests for a second chunk across 4,175 file-search tool responses in production agent logs, every one of which could have been paged

Agents don't paginate: the first chunk is the whole answer

When a tool returns more text than fits in an agent's turn, something gets cut, and the comfortable assumption is that the agent will ask for the rest. Across 4,175 file-search responses in production logs these authors saw not one request for a second chunk, so they measured the thing that follows from that: how often the file the agent needed was ranked first in the chunk it did get. A keyword scorer lifted that from 24.2% to 35.0%, and downstream the gain evaporated — the accuracy of naming the right file moved by at most 2.8 points across five models, none of it significant, because agents read the whole chunk and find the item wherever it sits. Spend your effort on what survives the cut rather than on ordering what already survived, and note that bolting four extra signals onto the keyword score — directory depth, file extension, recency, filename match — cost 4.8 points of first-place accuracy instead of adding any.

open the explainer → abstract PDF code

28% to 49%

share of each task's failing tests that the same model's patch fixes, once the harness trims older tool results instead of stalling when the context fills

Same model, different harness: different coding-agent results

A coding agent is a model plus a harness — the code that chooses what goes in the prompt, which tools exist, and when to stop — and the harness is usually treated as plumbing rather than as part of the thing being measured. Holding model, tasks and time limit fixed, this author kept the full transcript on disk but shortened older tool results to their opening and closing lines once the prompt passed half the window, and added a detector for repeated failing commands and for reading without editing. On 169 tasks with a 20,480-token window the mean share of a task's failing tests that got fixed went from 28% to 49% and outright solutions from 43 to 72; the same frozen settings lifted three other open-weight models by 6 to 20 points with no retuning. Widen the window to 262,144 tokens and the two arms tie at 69% versus 68.7%, so this is a fix for context pressure specifically — and it means a published agent score describes a model-and-harness pair, not a model.

open the explainer → abstract PDF code

0.95 to 1.10 times

the hallucination rate of abstention trained with no labels at all, measured against abstention trained on correctness labels, across six models

Can a model catch its own hallucinations for free?

Teaching a model to say "I'm not sure" normally starts with a labelled set: ask thousands of questions, mark which answers were wrong, fine-tune it to decline those. This paper drops the labels and ranks the training questions by the average log-probability the frozen model gave its own answer, retargets the least-confident slice to a fixed "I'm not sure." string, keeps the model's own answer everywhere else, and trains a small adapter. Across six open models from 1B to 8B that lands within noise of the label-supervised version at matched coverage — 0.95 to 1.10 times its hallucination rate, with overlapping intervals on every model — and the strongest of them drops from 44.5% of answers wrong to 16.4% across the 38% of questions it still takes on. The blind spot is the one you would fear: 36% of all questions come back wrong but confident, where the model's own doubt never fires, so treat this as a cheap first filter and keep retrieval or a second opinion behind it.

open the explainer → abstract PDF