GitHub Repo

GitHub 熱門專案、開源工具、框架、runtime 與重要 release。

Harness-of-Harness Extends Coding Agent Work Sessions Through Planning, Development, and Independent Testing Loops, but Its Core Implementation Is Not Yet Public

A team at the Shanghai AI Laboratory has introduced Harness-of-Harness, which adds a recurring control loop for planning, implementation, and verification around existing coding agents. Across three-round experiments, it delivered an average relative gain of 52.25%, but each condition used only a single generation, and the executable HoH-lite is still marked as coming soon.

2026-09-02

代理工具

Cline 4.1.17 Stops Resending Full Conversations to Clients and Closes a Checkpoint Commit-Overwrite Risk

Cline fixes an issue that broadcast the full transcript with every state update during long-running sessions, causing Hub memory usage to balloon to tens of gigabytes, and now sends only state snapshots. The release also tightens checkpoint handling, subagent cancellation, and remote MCP connection behavior, but deployments that do not pin models may produce different results after major default-model updates.

2026-09-02

RWKV7-G1j Scales an Attention-Free Recurrent Model to 13.3B While Maintaining Fixed State During Long-Sequence Decoding

The RWKV team has released the 13.3B-parameter G1j base model, along with Transformers, LoRA, and TileLang inference support. Its fixed-size recurrent state avoids a KV cache that grows with context length, but the model has no published safety or capability evaluations and should not be treated as a production-ready chat assistant.

2026-09-02

代理工具

Muse Code Goes Generally Available with Cross-Session Collaboration and Programmable Agent Control via MSP

Meta has taken Muse Code out of beta, adding cross-session messaging, sub-agent workflows, and confirmable rewind operations. The concurrently released TypeScript SDK exposes the Muse Session Protocol, but remains a pre-1.0 Developer Preview with no compatibility guarantees.

2026-09-02

TimesFM-3 Natively Supports Multivariate Zero-Shot Forecasting, but Its Weights Restrict Commercial and Production Use

Google Research’s 330-million-parameter TimesFM-3 can jointly forecast multiple time series in a single forward pass while incorporating historical and known-future covariates. Although the code is licensed under Apache 2.0, the new weights are separately subject to a noncommercial license, so production deployments cannot rely on assumptions carried over from previous versions.

2026-09-02

BenchMIRT Uses Multidimensional Item Response Theory to Dissect LLM Evaluations, Revealing That Safety Scores Can Conflate Reasoning Ability

Ai2 has open-sourced BenchMIRT, which separates latent capabilities such as safety and general reasoning based on model response patterns across more than 34,000 questions. Experiments show that some safety benchmarks primarily reflect reasoning performance and that a carefully selected 10% of questions can largely preserve capability rankings, though the data covers only models released by early 2025.

2026-09-02

代理工具

Gemini CLI 0.58 Blocks macOS Sandbox Access to Container Daemons, Closing Host Path That Could Bypass File Boundaries

The new release blocks sockets, executables, and IPC interfaces for container runtimes such as Docker, OrbStack, and Colima across all built-in Seatbelt profiles. The update also fixes write-policy configuration, the safety checker, and A2A cancellation state, but custom installation paths and existing workflows still require separate validation.

2026-09-02

推論系統

llama.cpp b10731 Adds Recurrent State Rollback for Qwen3.8-Flash-Next, Eliminating Repeated Full-State Transfers During MTP Speculative Decoding

llama.cpp fixes an issue in Qwen3.8-Flash-Next that prevented SSM and convolutional states from being rolled back correctly when draft tokens were rejected. Testing on a single RTX PRO 6000 showed code decoding performance rising from 123 to 183 tokens/s, though the results remain limited to a specific quantization, draft head, and single processing slot.

2026-09-02

推論系統

LMDeploy 0.17 Introduces DeepEPv2 and Mooncake KV Storage, Expanding Cross-Node MoE Inference Data Paths

LMDeploy 0.17.0 adds the DeepEPv2 expert-parallel backend to TurboMind and connects its KV connector to Mooncake Store. The release also supports Kimi K2.6, non-power-of-two page sizes, and structured responses, but provides no end-to-end performance data across hardware configurations.

2026-09-02

Vector databases and AI data infrastructure

LanceDB 0.38 Adds Incremental Materialized Views and Computed Columns, Alongside Multiple Breaking SDK Changes

LanceDB 0.38.0 brings materialized views, computed columns, and function resource requirements to Python, Node.js, and other interfaces, moving embedding and data transformation closer to the storage layer. The release also changes table detection, branch operations, and runtime requirements, so RAG systems should cross-check query results before upgrading.

2026-09-01

代理工具

Codex CLI 0.152 Adds Per-Tool MCP Output Quotas and Fixes Lost Instructions in Long-Running Sessions

OpenAI has added output token limits for individual MCP tools and now allows app-server shell commands to run for more than an hour. The release also fixes an issue where user instructions and authorizations could be lost after conversation history was compacted, while the built-in planning tool is now disabled by default.

2026-09-01

代理工具

VS Code Agent Host Moves the Agent Loop Out of Extensions, Using AHP to Synchronize Sessions Across Windows

VS Code 1.132 through 1.135 connect agents such as Copilot and Claude to a standalone Agent Host, allowing sessions to continue across windows, remote connections, and headless environments. The open Agent Host Protocol unifies session state, but not the reasoning, permissions, or tool behavior of individual models.

2026-09-01