Fetching from the wire…
01
02
03
04
05
06
07
08
Agents2026-08-20 · source-backed
A re-evaluation of two memory-bank self-improvement methods added two axes the original papers skipped: multiple runs to quantify variance, and randomly shuffled task order. Both exposed fragility. Reported gains depend heavily on the default task ordering, which acts as an implicit curriculum. (arXiv 2608.18066) Pair this with the WER finding above and a picture forms: the self-improvement literature has a measurement problem, and shuffling the inputs is a cheap way to find out if you have one too.
Each link below shares sources, entities, or timing with this story.
arXiv 2608.16402 defines security profiles and runtime obligations that compose through joins, intersections, budget narrowing, and approval inheritance. The useful part for builders is derivation: a sub-agent's effective policy comes from its parent's, rather than being hand-...
arXiv 2608.11879 benchmarked Mem0, Hindsight and Mastra Observational Memory across conversations up to 400 turns and 665 LoCoMo questions. Cost models built on conversation length miss badly because internal memory behavior dominates. Break-even against just replaying the ful...
MAFIA (arXiv 2608.03844) targets the two conditions that describe production and that prior attacks failed against: large benign memory pools and active input auditing. It adds placement strategy (probe memory, allocate injection budget, schedule writes to stay retrieval-compe...
"Memory in the Loop" (arXiv:2607.05690) moves memory read/write inside the agent's per-step loop, viable only with an in-process store answering in ~100µs. The behavioral number is the story: redundant actions were 0.0 of 12 at in-process speed but 7.2 of 12 at a 110ms cloud r...
Recuris (arXiv 2608.24876) keeps a Working Memory tracking current task progress separate from an Experiential Memory of learned skills, so skill selection indexes against what the task needs now rather than the whole history. It improves 35 of 37 model-benchmark pairs, gains...
Models navigate to the correct file for 92%+ of required deletions but cut the exact target line only 52% of the time, and 29% of passing patches wrap dead code in a conditional instead of removing it. Grep the diff for newly added if guards around code the task said to delete...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.