代理工具

Agent、MCP、coding agent、工作流自動化與評測。

EnvHarness Reshapes Agent Training Environments Without Modifying Verifiers, Raising Out-of-Domain ALFWorld Performance by 9 Points

Teams including Google Cloud AI Research apply the agent harness concept to the environment layer, using composable interfaces to alter initial states, visible information, and task flows. Open-source experiments span five benchmarks and deliver gains of up to 9 percentage points, although reproducing the full results still requires multiple external environments and commercial model APIs.

2026-08-22

代理工具

NVIDIA Launches CUDA MCP and Self-Hosted Nsight Blueprint, Keeping Four-Model Backend On-Premises

Nsight AI can now connect live CUDA documentation to MCP clients such as Codex and Claude, or deploy a complete coding-assistance backend locally. The open-source Blueprint combines generation, completion, embedding, and reranking models, but the reference configuration requires a Hopper-class GPU and at least 200GB of storage.

2026-08-21

Ornith-1.5 Brings Self-Generated Tasks Into Reinforcement Learning; 397B Model Self-Reports 86.1 on Terminal-Bench 2.1

Ornith has released three open-weight models: a 9B model, a 35B MoE, and a 397B MoE. The training loop jointly optimizes tasks, agent scaffolds, and solution trajectories. The 397B model’s coding-agent performance approaches that of closed-source frontier models, but the figures have not yet been independently reproduced, and the model’s release documentation has a licensing gap.

2026-08-21

代理工具

Claude Makes Computer Use, Skills, and the Files API Generally Available as Browser Agents Shift to Element-Level Control

Anthropic’s updated tools can perform multiple interface actions in a single model turn while combining page structure, versioned skills, and persistent files. Existing beta integrations will continue to work, but upgrading the computer use tool changes the request format, and engineering teams must still manage browser isolation and prompt-injection risks themselves.

2026-08-21

代理工具

NVIDIA AVO Completes All 183 Levels in the ARC-AGI-3 Public Set With About 12% Fewer Actions Than VISTA

NVIDIA adapted the AVO agent architecture, originally used for evolutionary search over GPU kernels, to interactive reasoning and achieved a 100.00 RHAE score with Claude Opus 5. The results show that agent memory and verification loops can substantially change the performance of the same model, but they cover only public environments, and the comparison with VISTA is not a controlled ablation.

2026-08-21

AI 安全

PolicyGuide Constrains Customer Service Agents with Persistent Workflow Graphs, Raising Four-Run All-Pass Rate from 0.42 to 0.62

PolicyGuide does more than block policy-violating actions before tool calls: it tracks incomplete policy steps during every user turn. It improved reliable success rates across three τ²-bench domains, but adds one LLM verification call per turn, and some procedural-compliance metrics were designed by the authors themselves.

2026-08-21

代理工具

Mojo Opens Compiler and Toolchain Source, Enabling Source Builds with a Single Bazel Command

One week after releasing Mojo 1.0, Modular has made the KGEN compiler, tools, and standard library available under Apache 2.0 with the LLVM exception. The source code can now be inspected and built independently, though the compiler is not yet accepting external contributions, and MAX remains under a separate license.

2026-08-21

代理工具

Agent Memory Transfers More Reliably When Broken Into Subtasks; Text Skills Lead Code by 2.9 Points on Average

A study spanning nine models and three long-horizon benchmarks finds that skills induced from complete tasks usually make agents perform worse, while those induced from subtasks deliver average gains. Natural-language process notes are also more robust than executable Python skills, though results remain highly dependent on the model and benchmark.

2026-08-21

DeepSeek Pro 再傳灰測：一句「I’m doing」讓社群聞到新模型氣味

8 月 21 日前後，DeepSeek Pro 灰測消息在開發者社群延燒；目前可確認的是官方 8 月 21 日公告只發布 Flash Vision 實驗模型，Pro 仍未公布新版本號。社群觀察到的「I’m doing」思維鏈與 B 站實測，更像一次小範圍路由或後端模型灰度放量。

2026-08-21

代理工具

MemTrapBench: Adding Relevant Memories Causes Five Agent Memory Strategies to Lose at Least 10.66 Points

MemTrapBench does not test whether memories can be retrieved. Instead, it examines whether correct, relevant past experiences can lock a model into the wrong strategy or belief. Across two models, all five memory methods underperformed the no-memory baseline, while a simple prompt-based guardrail recovered only part of the loss.

2026-08-21

代理工具

SAPO Lets a Single Autoregressive Model Serve as Both Actor and Critic, Speeding Up Each Agent RL Training Iteration by 33.2%

SAPO uses two causal boundaries in a generated sequence to read out state and action values simultaneously, eliminating the need for a separate critic or multiple rollouts of the same task. Qwen2.5-7B outperforms PPO and GRPO on the overall metrics for both ALFWorld and WebShop, although no public implementation is yet available for verification.

2026-08-21

推論系統

Axon Generates Five LLM Backends from a Strongly Typed DSL, Cutting MLX Median Latency to 48.3% of Transformers

Axon expresses model architectures as functional specifications with symbolic tensor shapes, then compiles them to PyTorch, Triton, JAX, MLX, and native vLLM. Across 467 inference tests, it accelerated most generative models, though some BF16, encoder, and sequence-to-sequence models still showed accuracy or performance regressions.

2026-08-21