Claude Code 2.1.259：managedMcpServers 与 headless prompt 模式
Tag v2.1.259 published 2026-09-02T22:33:51Z. Highlights: ## What's changed - Added `managedMcpServers` managed setting: organizations can provide HTTP/SSE MCP servers to every user (same entry shape as `.mcp.json`); entries that name a command to run are skipped - Added `--permission-prompts none` for unattended headless hosts: anything that would prompt is denied automatically while the active permission mode (including auto mode) keeps deciding - Added recognition of `glab mr create/merge/close/reopen/note/update` so GitLab merge requests show as `MR !N` in the collapsed tool summary and refresh the footer MR badge - Added `--json` to `claude plugin validate` for a machine-readable validation report - Fixed concurrent sessions silently reverting each other's `~/.claude.json` changes — workspace trust no longer resets and MCP/project state is no longer lost when running many sessions at once - Fixed a conversation whose thinking was rejected Stars about 143812.
microsoft/playwright-mcp published v0.0.80 at 2026-09-01T03:24:23Z (about 36742 stars). Adds browser_start_recording / browser_stop_recording (opt-in via --caps=devtools) that capture manual browser actions as Playwright code. Fixes: infer default --codegen language from environment; honor --user-data-dir in extension mode; browser_take_screenshot returns original bytes without silent downscale; drop cached browser backend after browser_close on shared context.
Anthropic公开页面claude.com/check-content提供「Check if a file was made with Claude」：用户在浏览器本地拖入JPG/PNG/SVG/视频/音频等（≤100MB），工具只读C2PA内容凭证、文件不离开设备。页面说明Claude对文本用水印、对支持文件类型附签名provenance；本工具仅识别文件凭证，不能判断文本是否由Claude撰写。文本水印检测API仍为面向监管/媒体/事实核查等机构的private preview。与Fable/Mythos 5.1水印承诺配套；HN 9/2讨论约149分。事件首次上线精确时刻未在页面标明。
Tag v0.59.0-preview.0 published 2026-09-01T20:19:07Z. Highlights: ## What's Changed * Changelog for v0.58.0-preview.0 by @gemini-cli-robot in https://github.com/google-gemini/gemini-cli/pull/29082 * chore(release): bump version to 0.59.0-nightly.20260825.g812f7a2bc by @gemini-cli-robot in https://github.com/google-gemini/gemini-cli/pull/29083 * fix(core): prevent SSRF in MCP OAuth metadata discovery and authentication by @josebalius in https://github.com/google-gemini/gemini-cli/pull/29081 * fix(core): enforce fail-closed workspace trust and filter mcpServers in restricted mode by @luisfelipe-alt in https://github.com/google-gemini/gemini-cli/pull/29099 **Full Changelog**: https://github.com/google-gemini/gemini-cli/compare/v0.58.0-preview.0...v0.59.0-preview.0 Stars about 106777.
展开编辑分析
关键事实：仍为 preview/prerelease（GitHub tag v0.59.0-preview.0）；正式稳定版待发。
为什么重要：Gemini coding CLI hardens MCP OAuth and trust boundaries
OpenExecutive 开源 AI 高管：8 个 Claude 专才代理，约 3476 stars
SenteLabsAI/OpenExecutive (README claims Apache 2.0; GitHub license field NOASSERTION) provides one executive persona backed by eight specialist agents (CSO/CFO/CHRO/GC/COO/CMO/CPO/Board) on FastAPI+Next.js+ChromaDB+SQLite episodic memory and a scheduler; defaults include claude-sonnet-4-6 with some roles on claude-opus-4-7. Repo created 2026-06-11; stars about 3476, forks about 332, pushed_at 2026-09-02. HN spiked about 1031 points on 2026-08-27 and resurfaced 2026-09-02. No GitHub Release tags. Multi-agent harness sample, not a model release.
Tag ai@7.0.91 published 2026-09-02T20:27:01Z. Highlights: ### Patch Changes - 802af1e: Add configurable recovery for provider errors received after `streamText` response streaming begins. Explicitly configuring `streamRetries` enables isolated retry attempts, including one bounded callback-directed recovery through `StreamTextOnErrorRetryCallback` with `streamRetries: 0`; recovered results and metadata reflect only the successful attempt, while the existing `StreamTextOnErrorCallback` contract and logging-only observer behavior remain compatible. - Updated dependencies [5484f27] - Updated dependencies [36eb7ee] - Updated dependencies [622fa7f] - @ai-sdk/gateway@4.0.73 Stars about 26552.
展开编辑分析
关键事实：GitHub release ai@7.0.91确认streamRetries相关变更。
为什么重要：Vercel weight-4: streaming agents can recover after provider errors mid-stream
Tag @e2b/python-sdk@2.46.4 published 2026-09-02T20:48:39Z. Highlights: ### Patch Changes - 6146cd0: Improve reliability for high-concurrency sandbox workloads by spreading envd traffic across four HTTP/2 connection pools. Set `E2B_ENVD_POOL_SHARDS` before importing the SDK to adjust the pool count. Stars about 0.
展开编辑分析
关键事实：SDK 包级星标未知；以 E2B 发行说明中的连接池改动为准。
为什么重要：Agent sandbox concurrency reliability; Cursor lists E2B as a Cloud Agent sandbox target
Tag v0.33.3-rc2 published 2026-09-02T00:11:33Z. Highlights: ## What's Changed * Report cached prompt tokens * Honor GGUF model defined default parameters * MLX, MLX-C, llama.cpp update ## New Contributors * @marcelpetrick made their first contribution in https://github.com/ollama/ollama/pull/17579 **Full Changelog**: https://github.com/ollama/ollama/compare/v0.33.2...v0.33.3-rc0 Stars about 180001.