Every processed story in chronological order, with the newest coverage first. Filter by tag, source, or score to drill in.
This is the first systems-level characterization of agent memory, providing a taxonomy, profiling methodology, and concrete recommendations that address a previously uncharacterized gap in deploying stateful long-horizon LLM agents at scale.
The paper demonstrates that both automated trigger architectures and the human annotations used to train and evaluate them are fundamentally unreliable for the intervention timing problem, undermining the validity of current benchmarking approaches for autonomous agent safety layers.
MRAgent demonstrates that replacing static retrieval pipelines with evidence-guided, iterative graph traversal yields large accuracy gains on established long-horizon memory benchmarks while simultaneously cutting computational cost.
SePO demonstrates that the prompt agent itself — not just the tasks it serves — can be a target of automated optimization, removing a hand-engineered bottleneck that prior prompt optimization methods left unaddressed.
The video demonstrates, with a concrete Terminal Bench result, that harness engineering can deliver large performance gains without any change to the underlying model — making it an accessible optimization path for practitioners who lack access to proprietary model fine-tuning.
The plugin centralizes a proven agentic text-editing pattern into a single reusable base, avoiding duplication across the multiple Datasette Agent plugins planned to depend on it.
The paper demonstrates that frontier CUA safety is domain-conditioned rather than general, meaning strong browser-surface defenses in Claude Sonnet 4.6 and GPT-5.4 do not extend to coding-agent contexts, and that published ASR benchmarks are unreproducible without the release of RL-optimized injection strings.
Cate represents a new entry in the open-source agentic coding IDE space, offering a canvas-based interface for coding workflows.
The post illustrates that MCP tool-catalog bloat can silently degrade Claude Code's tool selection accuracy, and that scoping servers to the project level and using a ranked-catalog gateway are concrete mitigations for the problem.
The post highlights a concrete security gap in MCP agent workflows — that a one-time tool approval does not account for subsequent changes to a tool's capability surface — and presents Interlock as an open-source mechanism to detect and quarantine such drift before execution.