# 火木工作室 / Fire Wood Studio > 专注 AI 原生产品的独立技术工作室。做桌面优先、本地可跑的 AI 工具:排版、推理、数据集、运维、流程各有专门产品。模型只是系统的一环,确定性逻辑留在代码里,数据留在用户自己的机器上。所有产品免费,无需注册账号即可下载。 > An independent studio building AI-native desktop tools. Local-first by design: deterministic logic stays in code, data stays on the user's own machine. All products are free to download and require no account. ## 站点 / Site - 中文首页 / Home (zh): https://tsagent.cc/ - English home: https://tsagent.cc/en/ - 博客 / Blog: https://tsagent.cc/blog/ · https://tsagent.cc/en/blog/ - 全部文章全文 / Full text corpus: https://tsagent.cc/llms-full.txt - RSS: https://tsagent.cc/rss.xml · https://tsagent.cc/en/rss.xml - Sitemap: https://tsagent.cc/sitemap.xml - 代码 / Code: https://github.com/MWang-TS - 联系 / Contact: support@tsagent.cc - 反馈 / Support: https://feedback.tsagent.cc/ ## 产品 / Products ### Easy Post - URL: https://ep.tsagent.cc/ · English: https://ep.tsagent.cc/en/ - Status: Live · Platforms: Windows, macOS - 简介 / Summary: 让社媒创作易如反掌 一款桌面内容生产工具:AI 灵感对话、素材整理、公众号排版、社交卡片生成,四个工作台覆盖从想法到发布的完整流程。排版由本地 Rust 引擎完成,不依赖大模型。 - Summary (en): Social content creation, made simple A desktop content studio with four workbenches — AI ideation, material organization, WeChat article formatting, and social card generation. Typesetting runs on a local Rust engine, not a model. - 主要功能 / Features: 灵感:AI 对话创作、素材:微信读书导入、图文:公众号排版、卡片:社交媒体信息卡片、模型配置自由、确定性引擎 ### Dr T - URL: https://drt.tsagent.cc/ · English: https://drt.tsagent.cc/en/ - Status: Live · Platforms: Windows, macOS - 简介 / Summary: Dr T,你的远程运维AI搭子 运维的高技术门槛被拆掉了:不用背命令,AI Agent 接管 SSH 自主排障;软件小巧,装在本机就能用。 - Summary (en): Dr T, your AI partner for remote ops The technical barrier to ops is gone: no commands to memorise, an AI agent takes over SSH and diagnoses on its own. A small client, installed locally. - 主要功能 / Features: 多设备 SSH 管理、Ask 模式:对话式诊断、Agent 模式:自主循环排障、本机模式:免 SSH 诊断、本地记忆存储、Skills 技能模板、报警接收:HTTP / MQTT、12+ 模型服务商,数据不出内网 ### Easy Infer Station - URL: https://eis.tsagent.cc/ · English: https://eis.tsagent.cc/en/ - Status: Live · Platforms: Windows, macOS, Linux - 简介 / Summary: 轻量化的 .pt 模型测试与演示工具 面向算法、产品与交付团队的桌面推理工具:加载 Ultralytics .pt 即可跑图片、批量、视频与 RTSP,不需要写代码也能做演示和效果评估。 - Summary (en): A lightweight tool for testing and demoing .pt models A desktop inference tool for algorithm, product, and delivery teams: load an Ultralytics .pt and run images, batches, video, or RTSP. Demos and evaluation without writing code. - 主要功能 / Features: 给客户演示、验证新模型、检测 + 分类、四种输入、ROI 与报警、跟踪与设备 - 源码 / Source: https://github.com/MWang-TS/easy-infer-station ### Easy Dataset - URL: https://eds.tsagent.cc/ · English: https://eds.tsagent.cc/en/ - Status: Live · Platforms: Windows, macOS, Linux - 简介 / Summary: 你的深度学习训练数据处理搭子 YOLO 数据集的本地工具箱:质量检测、带框可视化、图片标注(检测 / 旋转框 / 分割,支持 AI 预标注)、格式互转、划分合并、视频抽帧,从原始素材一路做到 data.yaml。 - Summary (en): Your sidekick for deep learning training data A local toolbox for YOLO datasets: quality checks, visual review, annotation (boxes, OBB, segmentation, with AI pre-labelling), format conversion, splitting and merging, and video frame extraction — from raw footage to data.yaml. - 主要功能 / Features: 数据集质量检测、带框可视化浏览、图片标注、AI 自动预标注、格式互转与导出、标签编辑、文件一致性管理、数据集划分、数据集合并、视频抽帧 - 源码 / Source: https://github.com/MWang-TS/easy-dataset ### Easy Conda - URL: https://easyconda.tsagent.cc/ · English: https://easyconda.tsagent.cc/en/ - Status: Live · Platforms: Windows, macOS, Linux - 简介 / Summary: 不用背 Conda 命令的环境管理器 轻量的 Conda 环境桌面管理器:自动发现已有的 Conda / Miniforge / Mamba,图形化完成环境创建、包安装与升级、渠道切换、environment.yml 导入导出、离线交付与缓存清理。 - Summary (en): The Conda manager that spares you the commands A lightweight desktop manager for the Conda you already have: discover Conda / Miniforge / Mamba, then create environments, install and upgrade packages, switch channels, import or export environment.yml, deliver offline, and clean caches — without typing commands. - 主要功能 / Features: 自动发现已有 Conda、Windows 与 WSL,一套界面、环境一屏看完,不用 conda env list、新建、克隆、删除都有向导、装包之前,先看求解结果、国内镜像一键添加、environment.yml 与精确清单、三种离线交付方式、一键激活并打开终端、日志、诊断与缓存清理 - 源码 / Source: https://github.com/MWang-TS/easy-conda ### Easy Trainer Station - URL: https://ets.tsagent.cc/ · English: https://ets.tsagent.cc/en/ - Status: In development · Platforms: Windows, macOS - 简介 / Summary: 把 YOLO 训练变成点几下的事 面向 YOLO 模型训练的桌面工作站(研发中)。图形化配置训练参数,实时查看 Loss / mAP 曲线,管理训练历史与模型版本,与 Easy Dataset 无缝衔接。 - Summary (en): YOLO training without the command line A desktop workstation for YOLO training (in development). Configure hyperparameters visually, watch Loss and mAP in real time, manage training history and model versions, and work directly with Easy Dataset. - 主要功能 / Features: 图形化配置训练超参数、实时 Loss / mAP 监控、训练历史与模型版本、与 Easy Dataset 联动 ### pt2rknn - URL: https://pt2rknn.tsagent.cc/ · English: https://pt2rknn.tsagent.cc/en/ - Status: Live · Platforms: Web - 简介 / Summary: PyTorch / ONNX 到 RKNN,三步搞定 面向 Rockchip NPU 的模型转换工具,Web 界面无需命令行。支持 YOLOv8 全系列(Det / Seg / Pose / OBB)、ResNet、RetinaFace 六种网络类型,覆盖 RK3562 到 RK3588,内置 INT8 量化与校准集自动构建,转换后用 Netron、电脑端基线推理和逐层误差分析当场验证。 - Summary (en): PyTorch and ONNX to RKNN in three steps A conversion tool for Rockchip NPUs with a web UI and no CLI. Six network types — the full YOLOv8 family (Det / Seg / Pose / OBB), ResNet, and RetinaFace — across RK3562 through RK3588, with INT8 quantization, an automatic calibration builder, and post-conversion checks via Netron, a desktop baseline, and per-layer error analysis. - 主要功能 / Features: 三步卡片式 UI、六种网络类型、双转换路径、INT8 量化 + 校准集自动构建、SSE 实时日志、转完当场验证、结果可追溯、WSL / Docker 双模式 - 源码 / Source: https://github.com/MWang-TS/pt2rknn ### GeniusFlow - URL: https://gf.tsagent.cc/ · English: https://gf.tsagent.cc/en/ - Status: Beta · Platforms: Web - 简介 / Summary: 把线下流程搬到线上,让 AI 参与每一步 面向企业的 AI 增强流程管理平台。业务人员用可视化设计器拖拽流程,每个节点内置专属 AI Agent,在执行阶段提供实时指引、知识推荐与自动质量检查。 - Summary (en): Digitize your process, with AI at every step An enterprise process platform with AI built in. Business users design workflows visually, and each node embeds an AI agent that guides execution, recommends knowledge, and checks quality automatically. - 主要功能 / Features: 可视化流程设计器、节点内 AI Agent、看板与甘特、多模型 + RAG 知识库、RBAC 权限、OpenAPI 中台模式 - 源码 / Source: https://github.com/MWang-TS/geniusflow ## 文章 · 中文 / Articles (zh) - [Personal Agent 调研报告:它是什么、现状如何、该怎么选](https://tsagent.cc/blog/personal-agent-what-is-it/): 一份关于个人代理(Personal Agent)的公开调研报告:用五个问题讲清它和聊天机器人、普通 Agent 的区别,五级主动性刻度,现有产品的真实水平与短板,四层能力的拼装方式,以及普通用户分四步逐步放权的选择建议。 - [我把自己的救援通道,一起搞挂了](https://tsagent.cc/blog/rescue-channel-on-the-broken-host/): 升级 openclaw 后重启失败,我照着报错里的建议停了网关服务——然后飞书 bot 一起哑了:我停掉的,正是一直给我出主意的那个 AI。这篇复盘一次很直白的单点故障,以及为什么真正救场的工具必须能独立于被修的那台机器存在。 - [把线下审批搬到线上:流程编排里 AI 该放在哪一步](https://tsagent.cc/blog/geniusflow-agent-workflow/): 流程数字化的失败,多数不是工具不行,而是把「判断」和「执行」混在了一起。这篇讲 GeniusFlow 的编排思路:哪些节点应该是确定性的规则、哪些适合交给 AI Agent、人工审批应该卡在什么位置。 - [pt / onnx 转 RKNN 上板:一份可照做的检查清单](https://tsagent.cc/blog/pt2rknn-quantization-checklist/): 转换成功不等于部署成功。这篇给出从 PyTorch / ONNX 到 RKNN 的完整检查顺序:模型结构选择、INT8 量化的校准集准备、逐层误差定位,以及上板后精度对不上的常见原因,适用于 RK3562 到 RK3588。 - [YOLO 训练怎么调参:先看清曲线,再动超参](https://tsagent.cc/blog/ets-yolo-training-params/): 训练效果不好时,多数人的第一反应是改学习率。其实更该先看图:Loss 震荡说明什么、mAP 提前平台意味着什么、什么时候该停。这篇讲 Easy Trainer Station 的观察方式,以及几条少走弯路的调参顺序。 - [Windows 上管 Conda 环境:不背命令,也不装 Docker](https://tsagent.cc/blog/easyconda-windows-gui-mirror/): conda 命令记不住、镜像源配错、环境换台机器就重建——这些事在 Windows 上尤其烦。这篇讲 Easy Conda 的用法:图形化建环境、切国内镜像、导入导出 environment.yml,以及怎么把环境打成离线包交付。 - [LabelMe 转 YOLO:格式转换之后,还有三件更该做的事](https://tsagent.cc/blog/eds-labelme-to-yolo/): 格式互转只是数据准备的第一步。这篇讲 Easy Dataset 的完整链路:LabelMe / VOC / YOLO / COCO 互转之外,还要做质量检测、标签一致性抽查与训练集划分,否则模型学到的可能是标注错误。 - [模型验收怎么做:用 RTSP 流和批量图片快速对比两个权重](https://tsagent.cc/blog/eis-model-acceptance-rtsp/): 训练完两个权重,光看 mAP 决定不了上不上线。这篇给出 Easy Infer Station 的验收流程:固定一组验证集跑批量对比,再用 RTSP 流看实时表现,把差异定位到具体场景,而不是只看一个平均分。 - [用自然语言排查一台远程服务器:Ask 与 Agent 两种用法](https://tsagent.cc/blog/drt-remote-ops-ask-agent/): 运维排查最耗时的是「看日志、猜原因、再验证」这三步循环。这篇讲 Dr T 的两种模式怎么配合:Ask 模式快速问清事实、Agent 模式连着做完一串动作,以及为什么关键命令仍然要你自己点头确认,模型只负责提速。 - [公众号排版:把排版交给本地引擎,把创意留给模型](https://tsagent.cc/blog/ep-wechat-formatting/): 很多工具让 AI 从标题到配图全包,结果文章读起来不像你写的。这篇讲 Easy Post 的做法:排版、编号、校验这类确定性工作交给本地 Rust 引擎,创意和最终判断留给自己,顺便给出一个可复用的写作流程。 - [AI 不是创作的主人,你才是 — Easy Post 的设计取舍](https://tsagent.cc/blog/ai-not-the-master/): 这篇文章讲我们做 Easy Post 时的取舍:排版交给本地 Rust 引擎保证确定性,创意交给模型,最终决定权留给你。也聊了为什么把 AI 塞进内容生产的每个环节,反而会让创作者失去对作品的控制,以及我们如何划清这条边界。 ## 文章 · English / Articles (en) - [Personal Agents: What They Are, Where They Stand, How to Choose](https://tsagent.cc/en/blog/personal-agent-what-is-it/): Personal agent research: how it differs from chatbots, a five-level autonomy scale, what today's products really do, and the four layers a full build needs. - [I took down my own rescue channel](https://tsagent.cc/en/blog/rescue-channel-on-the-broken-host/): The AI advising me ran on the gateway I was fixing. I stopped it, and lost my only way to ask anything. A postmortem of a very plain single point of failure. - [Moving approvals online: where AI belongs in a process](https://tsagent.cc/en/blog/geniusflow-agent-workflow/): Most process automation failures mix judgement with execution. How to classify nodes: deterministic rules, AI agent steps, and where approval belongs. - [PyTorch or ONNX to RKNN: a deployment checklist](https://tsagent.cc/en/blog/pt2rknn-quantization-checklist/): A successful conversion is not a working deployment: FP16 baseline, INT8 calibration data, per-layer error, and checks for when the board disagrees. - [Tuning YOLO training: read curves before hyperparameters](https://tsagent.cc/en/blog/ets-yolo-training-params/): Read the curves before touching hyperparameters: what a spiky loss means, what a flat mAP plateau means, and when to stop a YOLO training run. - [Managing Conda on Windows without commands or Docker](https://tsagent.cc/en/blog/easyconda-windows-gui-mirror/): Conda on Windows without memorizing commands: visual environments, China mirrors, environment.yml export and offline delivery, WSL included. - [LabelMe to YOLO: format conversion is the easy part](https://tsagent.cc/en/blog/eds-labelme-to-yolo/): Converting annotations is step one. The full path in Easy Dataset: format conversion, quality checks, label consistency and splits that do not leak. - [Accepting a YOLO model: batch images and a live RTSP stream](https://tsagent.cc/en/blog/eis-model-acceptance-rtsp/): Two checkpoints and mAP alone will not decide. Compare on a stratified batch, then watch an RTSP stream for instability a single score hides. - [Debugging a remote server: Ask mode vs Agent mode](https://tsagent.cc/en/blog/drt-remote-ops-ask-agent/): Most ops debugging is one loop: read, guess, verify. When to use Ask mode, when Agent mode, and why risky commands still stay with a human. - [WeChat formatting: let the local engine handle layout](https://tsagent.cc/en/blog/ep-wechat-formatting/): How Easy Post splits the work: layout, numbering and validation run in a local Rust engine, while ideas and final calls stay with you. - [AI is not the author — you are](https://tsagent.cc/en/blog/ai-not-the-master/): Why we put layout in a local engine, creativity in the model, and the final call in your hands — and how inserting AI into every step can cost creators control. ## 引用说明 / Citation 引用本站内容时请注明来源「火木工作室 / Fire Wood Studio」并链接到对应页面。When citing, attribute to "Fire Wood Studio" and link to the source page.