Archive
Agent、MCP、coding agent、工作流自動化與評測。
AI 安全
Researchers used Python module shadowing to make a security decoder written by Claude Code load a malicious `struct.py`, ultimately achieving remote code execution. The case shows that per-tool-call classifiers cannot replace operating system isolation, and that a zero percent success rate in official evaluations cannot be extrapolated to unseen attack chains.
代理工具
The Model Hardware Standard packages instrument capabilities, natural-language descriptions, and safety constraints into discoverable driver interfaces, enabling agents to coordinate multiple devices through MCP, a CLI, or APIs. It remains an application-only research preview, with the specification, code, and cross-model safety evaluations not yet public.
代理工具
Cursor no longer requires users to connect an existing GitHub or other source code repository first. Agents can create projects directly and operate live environments through the browser. The new workflow shortens the path from prototype to deployment, while expanding the audit scope for cloud execution, network exposure, and software supply chain permissions.
AI 研究
CritICL first catalogs small-model errors, failure types, and critiques offline, then adds relevant examples to a large model’s prompt instead of repeatedly sampling for every question. The authors report slightly higher accuracy than self-consistency on Qwen2.5 math benchmarks, but the cost of building the library offline and the unreleased code limit reproducibility.
代理工具
Experiential unifies cloud, bring-your-own-key, and local models behind an OpenAI-compatible interface, with identity scoping, model allowlists, and hard budget controls. It can also build task routers from OpenTelemetry agent traces, although its cost-savings figures currently come primarily from the team’s own evaluations.
GitHub Repo
Debian developers selected the “Responsible Use of Generative AI” proposal, retaining existing quality and licensing standards and holding submitters fully responsible for AI output. The resolution does not mandate disclosure of AI use, but explicitly restricts sending sensitive data to external services and requires large-scale automated changes to be discussed in advance.
代理工具
The new release adds the `on_mcp_tool_result` lifecycle interface, allowing plugins to inspect or replace successful and error results from MCP. This provides a new control point for data masking and output normalization, while also elevating plugins into an agent trust boundary that requires auditing.
代理工具
The OpenAI Python SDK has switched its default synchronous and asynchronous clients from HTTPX to HTTPX2 and no longer includes the legacy `httpx` and `certifi` packages as transitive dependencies. General API calls remain largely unchanged, but enterprise proxies, custom transports, RESPX tests, and certificate trust configurations may break after upgrading.
代理工具
PILOT connects a supervisor and a worker agent through a bidirectional channel, allowing the former to provide guidance or terminate an erroneous path before a long-running task is complete. The authors report improved accuracy and token efficiency across three agent benchmarks, but its “self-improvement” updates only the harness, skills, and memory—not the model parameters.
最新模型
CLAP first learns physical dynamics from unlabeled videos through latent actions, then maps the representation back to robot end-effector coordinates. The research team has released the code and multiple checkpoints, but physical hallucinations in generated videos still limit its direct use in safety-critical control.
代理工具
PRAXIST has multiple research agents propose changes over successive generations, then records evaluation results, mechanisms, and lineage in an auditable evidence graph to avoid restarting exploration in every round. Its authors report 60 medals across 75 MLE-bench tasks at a model cost of $3,054, but the newly released system’s performance and costs have so far been measured primarily by its development team.
最新模型
LeVJEPA uses a single encoder, a cross-view invariance loss, and SIGReg to prevent representation collapse, eliminating the need for an EMA target network, stop-gradient, or a pixel decoder. Under the same data and training schedule, the authors report a 5.6× to 20.8× compute difference, though the results remain limited to frozen-encoder benchmarks based on models retrained by the team.