Xskill

Latest version: v0.6.26

Safety actively analyzes 971914 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

Page 1 of 5

0.6.26

紧急修复:sweep 按任务完成顺序及时回写状态,停服与超时后显式回收子进程;暂停用户尚未完成拆分的轨迹不再进入面板展示和指标计算,恢复后自动重新纳入。

0.6.25

What's Changed
* release: prepare xskill v0.6.25 by 370025263 in https://github.com/SkillNerds/xskill/pull/129


**Full Changelog**: https://github.com/SkillNerds/xskill/compare/v0.6.24...v0.6.25

0.6.21

概要 / TL;DR

修复线上 team-server 在冷启动/高负载下**反复卡死重启**的问题,并把架构改成"重计算与前台隔离",在**万级 skill / 十万级轨迹**下也撑得住。

Fixes the team-server **repeated freeze-and-restart** under cold-start / heavy load, and re-architects heavy compute to run isolated from the web process — holds up at **10k+ skills / 60k+ trajectories**.

> ⚠️ 升级后请盯一下第一波冷启动:探针不再超时、`py-spy` 里不再有 `find_atom_entry_in_any_skill` 热栈即为正常。

根因 / Root cause

现象:`health` / `watcher` / `web` 三个探针**同时** 5s 超时,进程被反复重启。`py-spy` 定位:watcher 热循环 `find_atom_entry_in_any_skill` 对每个未打 `clustered` 标记的 atom **全量重读所有 skill 的 `.candidates.yml`**(O(atoms×skills)),烧满一个核、握死 GIL,饿死跑异步探针的事件循环。embedding 后端其实很快,**并非瓶颈**。

py-spy showed the watcher hot loop re-reading every skill's `.candidates.yml` per un-clustered atom (O(atoms×skills)), pegging one core and starving the async event loop via the GIL. Embedding was not the bottleneck.

主要改动 / Highlights

- **热循环惰性缓存**:每轮 sweep 惰性构建一次 `atom_id→skill` 索引,仅在遇到未打标 atom 时才建;稳态全 `clustered` **零扫盘**(1 万 skill 不再每轮白扫)。O(atoms×skills) → O(skills)。
- **watcher / 画像拆为定时短命子进程**:web 进程只留轻量调度线程 spawn 子进程,重计算全在独立解释器进程,**GIL 与事件循环隔离**;算完即退,不留常驻线程。
- **前端海量 skill 分页**:`/api/v1/dashboard/skills` 加 `limit`/`offset` 分页 + `name` 定向查,前端翻页,1 万 skill **不炸锅**。
- **看板随主 app 挂载**(主进程轻量后无需独立进程);观测端点 `/watcher/status`、`/stats` 改读子进程落盘状态文件。
- **画像同步默认放慢**到 600s(10 分钟);新增 `server.sweep_timeout` / `profile_refresh_interval` / `profile_refresh_timeout` 配置。

升级 / Upgrade

bash
pip install -U xskill → 0.6.21
重启:xskill serve --server(自动挂看板 + 起 sweep/profile 内部调度器)


测试 / Tests

全套 **1573 单测通过**。唯一未过的是 e2e `test_canary_flip_promote_and_install_new_version` —— 系环境问题(该 e2e 跑机器上**已装的 0.6.16 二进制**而非源码,而环境 `websockets 16` 移除了 `WebSocketServerProtocol` 打挂了 0.6.16),与本次改动无关。

0.6.20

新功能:`xskill init`

一条命令上手:把官方 **xskill 使用指南 skill** 装进你机器上所有 agent 生态(Claude Code / Codex / OpenCode / Cursor / Trae…),再交互式问 server / token / 工号连上团队 server。


xskill init 全交互
xskill init 10.0.0.5:8000 --token <t> --name <工号> --yes 无头
xskill init --skills-only 只装指南 skill

已有常驻连接时会展示当前状态并询问是否停掉重配(`--force` 直接替换)。装好后在任意 agent 里 `/xskill` 即可查升级/调试/搜索/各平台用法。

修复:清掉剩余的 Windows 弹窗面

0.6.19

紧急修复:Windows 无限弹 cmd 黑窗(125, PR 126)

**强烈建议所有 Windows 用户立即升级**:`pip install -U xskill`

0.6.18.post2

What's Changed
* chore(sse): 注释停用三个无调用方的 SSE 流水线端点 by 370025263 in https://github.com/SkillNerds/xskill/pull/105
* perf(skillhub): 三层扫描缓存,根治全盘扫描风暴(面板冻结事故修复) by 370025263 in https://github.com/SkillNerds/xskill/pull/108
* refactor(cli): search 只保留 skillhub 形态 + 网络异常兜底 + 来源/评分展示 by 370025263 in https://github.com/SkillNerds/xskill/pull/107
* fix(updater): server 请求默认直连 + pip 升级闭环核验与回退(内网镜像滞后事故) by 370025263 in https://github.com/SkillNerds/xskill/pull/109
* feat(skillhub): search 升级 BM25+语义 RRF 混合检索(含 embedding.max_embed 护栏) by 370025263 in https://github.com/SkillNerds/xskill/pull/110
* perf(dashboard): 画像散点事件触发物化 + multiprocessing 隔离(closes 106) by 370025263 in https://github.com/SkillNerds/xskill/pull/111
* fix(skillhub): 收紧搜索 embedding 并修复技能包分发 by 370025263 in https://github.com/SkillNerds/xskill/pull/113
* fix(scatter): 停机时取消排队计算并等待派发线程 by 370025263 in https://github.com/SkillNerds/xskill/pull/114
* fix: clean owned search slot installations by 370025263 in https://github.com/SkillNerds/xskill/pull/116
* fix(skillhub): keep degraded search fast by 370025263 in https://github.com/SkillNerds/xskill/pull/115
* test(release): gate SkillHub workloads and harden SQLite close by 370025263 in https://github.com/SkillNerds/xskill/pull/117
* fix(release): stabilize pre-tag validation gates by 370025263 in https://github.com/SkillNerds/xskill/pull/118
* perf(skillhub): make release smoke truly cache-hot by 370025263 in https://github.com/SkillNerds/xskill/pull/119
* perf(skillhub): cache hot hybrid rank plans by 370025263 in https://github.com/SkillNerds/xskill/pull/120
* test(release): refresh hot snapshots before timing by 370025263 in https://github.com/SkillNerds/xskill/pull/121
* perf(skillhub): cache hot search results by 370025263 in https://github.com/SkillNerds/xskill/pull/122
* perf(skillhub): isolate hot search requests by 370025263 in https://github.com/SkillNerds/xskill/pull/123
* perf(skillhub): yield during hot search bursts by 370025263 in https://github.com/SkillNerds/xskill/pull/124


**Full Changelog**: https://github.com/SkillNerds/xskill/compare/v0.6.17...v0.6.18.post2

Page 1 of 5

© 2026 Safety CLI Cybersecurity Inc. All Rights Reserved.