20:"$Sreact.view_transition" 21:I[9754,["/_next/static/chunks/2t36f5p4pjzlb.js","/_next/static/chunks/1gyqeidjm5eoa.js","/_next/static/chunks/0_fjnxc4tt_l_.js","/_next/static/chunks/1ymt1shyhmu-4.js","/_next/static/chunks/3kv2zf0i7s2wc.js","/_next/static/chunks/1vs_3g_g75zmk.js","/_next/static/chunks/2g4dh802s8oz3.js","/_next/static/chunks/1bc3f0shukf1j.js","/_next/static/chunks/192kgs9v_t14j.js","/_next/static/chunks/0z6pw0v5c_lx5.js","/_next/static/chunks/08r60e6icmtkc.js","/_next/static/chunks/3trx6c0i7_x-3.js","/_next/static/chunks/40sro5bzp9bi_.js","/_next/static/chunks/1wgf3tk38y6cf.js"],"ProjectDetail"] 22:T4e52, Serena · 编码代理的符号级语义 IDE

Serena · MCP toolkit

让你的编码代理沿着符号的轨迹，
在代码库里精确飞行

Serena 是一个通过模型上下文协议（MCP）工作的工具包，为 AI 编码代理提供符号级的语义检索、编辑与重构。它把代码库当作一张可导航的关系星空，代理沿着引用轨道定位答案，而不是在文件行号里盲目搜索。

symbol-level retrieval 40+ languages cross-file refactor memory system

查看 GitHub 仓库 oraios/serena

01 / RETRIEVAL

find-references

按符号定位答案，而非按行号碰运气

传统的代码检索按文件与行号组织，代理只能一次次猜测再验证。Serena 直接在符号层面理解代码：函数、类、变量的定义与引用关系都是一颗颗星、一条条星轨。

当你问「这个函数被谁调用」，它给出的不是一堆文件，而是围绕目标符号展开的引用星图——每个节点就是一次调用点，每条切线标注着 find-references 关系。

» def build_index( ... ) — 引用：14

02 / REFACTORING

cross-file

跨越文件边界的重构，像一次重新布线

重构最危险的地方在于承重结构：改掉一个符号，可能牵动几十个文件。Serena 在动手前先读懂全部的引用关系，把跨文件的依赖当作一条条需要同步重连的星轨。

它支持重命名、移动、内联等操作，基于符号级理解保证改一处、处处同步——而不是在每个文件里单独修补碰运气。变化的范围在动手前便以关系图呈现，你可以先看清再确认。

部分高级重构能力通过 JetBrains 插件提供（付费，含免费试用）。

03 / SYMBOLIC EDITING

signatures

在符号边界上编辑，token 用得更省

面对一个大型函数，代理往往需要把整块代码读进来才能修改一行。Serena 提供符号级编辑工具：替换某个符号的实体、在其前后插入内容，都直接以符号为单位完成。

这样既减少了上下文消耗，也让编辑请求更精确——改的是「这个符号」，而不是「这一段文本」。符号的边界、引用与签名在一次操作中被同时理解。

» edit_symbol(body=Pagination.parse, action=replace)

04 / MEMORY

cross-session

记忆把知识留在星图里，跨会话累积

好的代理不该每次都从零认识你的代码库。Serena 内置记忆系统，在会话之间、用户之间、项目之间共享那些你已经确认过的结构理解。

这些积累下来的知识如同星轨分支，指向一层记忆光晕：下次再进来，代理已经知道哪些符号是关键路径、哪些约定要保持。所有行为都通过多层级 YAML 配置灵活裁剪，适应不同的团队口径。

05 / LANGUAGES

grammars behind

一套星图，覆盖 40+ 种语言

Serena 通过语言服务器后端（language server backend）理解不同语言的符号结构，超过 40 种编程语言共享同一种语义星图。语言只是不同的语法表，符号与引用的关系在各类语言间保持一致。

40+

识别范围以语言服务器后端实际支持为准。

pythontypescriptrustgojava c#c++rubyphpkotlin swiftc…

06 / INTEGRATE

MCP everywhere

因为一切通过模型上下文协议（MCP）展开，Serena 不绑定单一客户端。它接入 Claude Code、Codex、OpenCode、Gemini-CLI 等终端客户端，也有面向 VSCode、Cursor、JetBrains 的 IDE 集成方式。技术专名保持原名，便于你在自己的工具链里直接对应。

Claude CodeCodexOpenCodeGemini-CLI VSCodeCursorJetBrains

23:T4a8f, Serena · A symbol-level semantic IDE for your coding agent

Serena · MCP toolkit

Let your coding agent
fly along the symbols,
not gamble on line numbers

Serena is an MCP-based toolkit that gives AI coding agents symbol-level semantic retrieval, editing and refactoring. It treats your codebase as a navigable constellation, letting agents travel the reference orbits — instead of searching blindly through files.

symbol-level retrieval 40+ languages cross-file refactor memory system

Visit the GitHub repo oraios/serena

01 / RETRIEVAL

find-references

Locate by symbol, not by line number

Classic code search is organized around files and line numbers, so an agent keeps guessing and re-checking. Serena understands your code at the symbol level: functions, classes and variables each become a star, and every definition-to-reference edge becomes an orbit.

Ask "who calls this function" and you get more than a list of files — you get a reference constellation around that symbol, where each node is a call site and each tangent is a find-references relationship.

» def build_index( ... ) — references: 14

02 / REFACTORING

cross-file

Refactoring across files, like rewiring a circuit

The danger in any refactor is the load-bearing structure — rename one symbol and dozens of files may depend on it. Serena reads the full reference graph before it acts, treating every cross-file dependency as an orbit that must be reconnected together.

Rename, move and inline operations are driven by symbol-level understanding, so a change lands consistently everywhere at once — rather than being patched file by file and hoping nothing breaks. The blast radius is shown as a relationship map up front, so you can review before you commit.

Some advanced refactors are available through the JetBrains plugin (paid, with a free trial).

03 / SYMBOLIC EDITING

signatures

Edit at symbol boundaries, spend fewer tokens

Faced with a large function, an agent often has to pull in the whole block just to change one line. Serena ships symbol-level editing tools: replace a symbol's body, or insert content before and after it, all addressed as a unit rather than a stretch of text.

That cuts context consumption and makes the request more precise — you're editing "this symbol", not "this paragraph". The symbol's boundaries, references and signature are understood together within a single operation.

» edit_symbol(body=Pagination.parse, action=replace)

04 / MEMORY

cross-session

Memory keeps knowledge in the map, session to session

A good agent shouldn't re-learn your codebase from scratch every time. Serena ships a memory system that shares structure you've already confirmed across sessions, users and projects.

These accumulated insights sit like orbit branches pointing toward a memory halo: on the next run, the agent already knows which symbols sit on critical paths and which conventions to respect. Everything is trimmed through multi-layered YAML configuration, so each team can shape it to its own standards.

05 / LANGUAGES

grammars behind

One constellation, 40+ languages

Serena understands symbol structure across different languages through a language server backend, letting more than 40 programming languages share the same semantic constellation. A language is just a different grammar — the symbol-to-reference shapes stay consistent everywhere.

40+

Coverage depends on what the language server backend actually supports.

pythontypescriptrustgojava c#c++rubyphpkotlin swiftc…

06 / INTEGRATE

MCP everywhere

Because everything runs over the Model Context Protocol (MCP), Serena isn't tied to a single client. It plugs into terminal clients like Claude Code, Codex, OpenCode and Gemini-CLI, and there are IDE integrations for VSCode, Cursor and JetBrains. We keep the product names in their original form so you can match them directly against your own toolchain.

Claude CodeCodexOpenCodeGemini-CLI VSCodeCursorJetBrains

a:[["$","$20",null,{"enter":{"secondary-forward":"secondary-forward","secondary-back":"secondary-back-enter","default":"none"},"exit":{"secondary-forward":"secondary-forward","secondary-back":"secondary-back-exit","default":"none"},"update":{"secondary-forward":"secondary-forward","secondary-back":"secondary-back-update","default":"none"},"default":"none","children":["$","$L21","cmsn44dhj00281s164ac5vy5q:zh-CN:beautified",{"project":{"id":"cmsn44dhj00281s164ac5vy5q","slug":"oraios-serena","name":"serena","description":"为 AI 代理提供语义代码检索与编辑的 MCP 工具包","authorName":"oraios","authorVerification":{"verified":false,"verifiedAt":null},"category":"development_tools","subcategory":"data_ai_tools","experienceUrl":"https://oraios.github.io/serena","qualityScore":100,"featured":false,"officialSite":null,"views":{"original":{"available":true},"official":{"available":false,"reason":"disabled_by_policy"},"beautified":{"available":true,"reason":null},"defaultView":"beautified"},"beautifiedView":{"revisionId":"cmt2zh25401ap62715suub1gr","kind":"artifact","artifact":{"schemaVersion":"localized-static-html-v4","kind":"static_html","locales":["zh-CN","en"],"defaultLocale":"zh-CN","documents":{"zh-CN":"$22","en":"$23"},"designDirections":{"schemaVersion":"dodot-design-directions-v1","directions":[{"id":"symbol-constellation","name":"符号星座","projectMetaphor":"把代码库当作一张可导航的关系星空图：每个符号（函数/类/变量）是一颗星，引用关系是连接它们的星轨。Serena 让代理沿着这些语义轨道精确飞行，而不是在文件行号里盲目搜索。","signatureVisual":"符号星座图：以当前符号为中心的光点，向被引用/引用它的符号辐射切线，切线标注 find-references 关系；无终端、无编辑器壳、无网格背景。","heroGeometry":"居中发光中心符号点，多条向四周放射的弧线轨道连接次级符号点，弧线带渐变色描边与可移动的流动光点。","sectionSequence":["constellation-hero","symbol-retrieval","cross-file-refactoring","signature-editing","memory-system","language-support"],"surfaceModel":"深空色底 + 不同透明度叠加的星层表面；每层以微弱光晕而非实色边框分隔，悬浮卡片为“星群”。","palette":"深空底 #070a13、星层表面 #0e1424、墨 #e7ecf5、主 #6ea8ff、引用关系 #c084fc、弱化 #8b98b8","typography":"无衬线叙事正文 + 等宽标注符号名称与语言标识；光点标签用等宽小号。","desktopComposition":"Hero 居中大坐标系：左上是符号检索输入示意，右上是星座主视觉，下方流程章节按检索→重构→编辑→记忆→语言依次铺开。","mobileTransformation":"星座收窄为单列竖向星轨时间线，两侧符号点折叠成垂直引用列表，每节标题前置一颗标记符号。","mediaStrategy":"无本地合格素材，删除媒体区；全部为 derived_visual 关系示意，不伪造截图。","forbiddenPatterns":["浏览器/手机/终端/编辑器壳","三张指标卡片","发光网格或粒子墙面","伪造产品截图"],"compositionFingerprint":"放射星轨星座，中心符号主导，弧线关系连线贯穿六节"},{"id":"construction-bid","name":"语义施工图","projectMetaphor":"把一次重构成一次带标注的工程蓝图：符号是待加工的构件，引用是承重关系，虚线裁剪线表示安全移除，实线箭头表示新连接。Serena 在动手前先读懂承重结构。","signatureVisual":"符号蓝图局部：代码符号以抽象构件块呈现，方框角标标注符号类型，引用以虚/实折线连接，移除线为红色虚线带剪刀标记，新增线为实线。","heroGeometry":"棋盘式的横向施工分解轴：若干符号构件沿 x 轴排布，折线引用在构件之间穿梭，下方一条裁剪示意带。","sectionSequence":["blueprint-hero","load-bearing-references","safe-removal","reconnect-refactor","symbol-editing","memory-system","language-support"],"surfaceModel":"图纸式浅底纸面：细分规格标注、边框 1.5px 实线细分样式、对角标注线；表面为带规格网格的工程纸质感（非发光网格）。","palette":"图纸浅底 #f4f5f3、纸面 #ffffff、墨 #1b1e22、主 #b5375a、移除线 #c0392b、新连线 #2563a8、弱化 #6b7480","typography":"等宽结构标题 + 工程标注标签；说明性文案用紧凑无衬线。","desktopComposition":"Hero 为横向施工轴主视觉，左右对称放置新旧引用标注；下方章节按“承重评估→安全移除→重连→编辑→记忆→语言”纵向承接，图纸标注贯穿。","mobileTransformation":"横向施工轴纵向堆叠为分层剖面图，构件块上下叠放，引用用左右弯折线；四节以上在窄屏压缩为折叠式规格表。","mediaStrategy":"无本地合格素材，删除媒体区；全部为 derived_visual 蓝图式关系示意。","forbiddenPatterns":["浏览器/手机/终端/编辑器壳","三张指标卡片","发光网格或粒子墙面","伪造产品截图"],"compositionFingerprint":"工程图纸施工轴，构件块 + 虚/实折线引用 + 红色移除虚线"},{"id":"retrieval-vector-field","name":"检索流场","projectMetaphor":"把一次语义检索当作被查询的含义在语言空间中激起的一串波场：查询向量像投入代码库的一颗粒子，符号按语义相关性沿流线被吸附到圆周上，越相关离核心越近。Serena 让代理按语义而非文件位置定位答案。","signatureVisual":"语义流场截面：交互式查询提示语写在左侧，右侧是同心圆场，符号点沿语义流线被吸聚到距圆心不同半径的轨道上，半径距离即相关度；无终端、无编辑器壳、无网格背景。","heroGeometry":"居中偏右的同心圆辐射场，查询提示语从左侧水平射入核心，相关符号点沿螺旋流线分层吸附在外环，最相关者贴近圆心。","sectionSequence":["vectorfield-hero","semantic-ranking","symbol-retrieval","symbolic-editing","refactoring","memory-system","language-support"],"surfaceModel":"深底但偏蓝紫的连续调子层：同心圆为透明度渐变环，节与节以波层（而非硬边框）过渡；表面高度依赖圆的层级嵌套。","palette":"深底偏蓝紫 #0a0d1a、连续层 #121628、墨 #eef0fb、主 #3f8cff、相关度 #2fd4b4、弱化 #9aa3d0","typography":"无衬线叙述 + 查询词等宽前置；相关度用下标等宽数值标注。","desktopComposition":"Hero 为左查询、右同心圆的横向流场；下方章节以“相关排序→检索→编辑→重构→记忆→语言”分双栏流线继续铺开。","mobileTransformation":"同心圆场压缩为单颗核心 + 环形列表：最相关符号置顶，其余按环序竖向排列，查询词固定为页首一行。","mediaStrategy":"无本地合格素材，删除媒体区；全部为 derived_visual 关系示意。","forbiddenPatterns":["浏览器/手机/终端/编辑器壳","三张指标卡片","发光网格或粒子墙面","伪造产品截图"],"compositionFingerprint":"左射入查询 + 同心圆语义流场，半径即相关度的吸附关系"}],"selectedDirectionId":"symbol-constellation","selectionRationale":"Serena 的核心机制是符号级语义检索与引用关系导航，而非文本替换或图形化重构演示。三个方向里，检索流场以“相关度同心圆”表达检索语义权重，施工图以“剪切/重连蓝图”表达重构，符号星座则以引用关系（find-references）为唯一主视觉。符号星座在检索、重构、编辑、记忆、语言六个章节上都能自然延续同一种关系视觉，且完全避开历史种子里的编辑器/终端壳与营销 Hero/指标卡默认；检索流场和施工图分别偏重单轴（检索、重构），对多能力全景的承载较弱。综合证据契合度与六节构图连续性，选择符号星座。"},"designBrief":{"schemaVersion":"dodot-design-brief-v1","selectedDirectionId":"symbol-constellation","projectMetaphor":"把代码库当作一张可导航的关系星空图：每个符号（函数/类/变量）是一颗星，引用关系是连接它们的星轨。Serena 让代理沿着这些语义轨道精确飞行，而不是在文件行号里盲目搜索。","signatureVisual":"符号星座图：以当前符号为中心的光点，向被引用/引用它的符号辐射切线，切线标注 find-references 关系；无终端、无编辑器壳、无网格背景。","compositionFingerprint":"放射星轨星座，中心符号主导，弧线关系连线贯穿六节","heroGeometry":"居中发光中心符号点，多条向四周放射的弧线轨道连接次级符号点，弧线带渐变色描边与可移动的流动光点。","sectionSequence":["constellation-hero","symbol-retrieval","cross-file-refactoring","signature-editing","memory-system","language-support"],"surfaceModel":"深空色底 + 不同透明度叠加的星层表面；每层以微弱光晕而非实色边框分隔，悬浮卡片为“星群”。","palette":"深空底 #070a13、星层表面 #0e1424、墨 #e7ecf5、主 #6ea8ff、引用关系 #c084fc、弱化 #8b98b8","typography":"无衬线叙事正文 + 等宽标注符号名称与语言标识；光点标签用等宽小号。","desktopComposition":"Hero 居中大坐标系：左上是符号检索输入示意，右上是星座主视觉，下方流程章节按检索→重构→编辑→记忆→语言依次铺开。","mobileTransformation":"星座收窄为单列竖向星轨时间线，两侧符号点折叠成垂直引用列表，每节标题前置一颗标记符号。","mediaStrategy":"无本地合格素材，删除媒体区；全部为 derived_visual 关系示意，不伪造截图。","evidencePlan":["符号级检索：以中心符号辐射 find-references 切线，表达符号级理解机制","跨文件重构：弧线轨道跨越章节表示跨文件引用关系，承托“跨文件重构”声明","40+ 语言支持：等宽语言标识光点绕轨道排布，以关系表达语言范围而非表格","记忆系统：星轨分支指向记忆层光晕，表示跨会话知识共享"],"localeContract":{"zh-CN":"为你的编码代理提供 IDE 级能力——符号级语义检索与编辑的 MCP 工具包，支持 40+ 语言与跨会话记忆。","en":"IDE-level capabilities for your coding agent — an MCP toolkit for symbol-level semantic retrieval and editing, 40+ languages, and cross-session memory."},"forbiddenPatterns":["浏览器/手机/终端/编辑器壳","三张指标卡片","发光网格或粒子墙面","伪造产品截图","重复历史种子中的编辑器面板 + 引用连线构图"],"historicalReuse":{"sourceAssetId":"HEAD:deepseek-v1/oraios-serena","decision":"repair","rationale":"历史种子把主要视觉锚定在“编辑器面板 + find-references 连线”，既与 antiTemplateContext 无关但在机制表达上偏工具壳。本案保留其“符号级 / 引用关系”这一核心想法的构图种子，但重构为无壳的星系轨道形式，摆脱编辑器轮廓，避免直接沿用旧 HTML。","reusedElements":["符号级代码理解作为视觉主题","find-references 引用连线作为关系语言","深色代码风格色系基调（重构为深空底）"],"discardedElements":["编辑器/IDE 面板壳（工具轮廓）","三栏（工具面板+代码面板+侧栏）桌面构图","等宽为主的全文排版","旧 v1 schema 与字段"]}},"visualReview":{"schemaVersion":"dodot-automated-design-review-v1","status":"approved","summary":"页面忠实实现 design-brief.json 选定的 symbol-constellation 方向：居中发光中心符号点配合放射切线轨道（find-references 关系）贯穿六节，配色与 brief 指定深空色板一致，视觉锚点为项目专属星座图而非通用 Hero 或三卡。所有能力声明均有 material.json 材料依据，未出现编辑器/终端壳、三卡指标墙、伪造截图或假控件。","checks":{"projectFit":{"passed":true,"evidence":"六节顺序 symbol-retrieval / cross-file-refactoring / signature-editing / memory-system / language-support（另含 integrators）与 brief.sectionSequence 一致；色板 --bg:#070a13、--layer:#0e1424、--ink:#e7ecf5、--primary:#6ea8ff、--ref:#c084fc 全部命中 brief.palette；sans 正文 + 等宽符号标注符合 typography 策略。"},"evidenceBound":{"passed":true,"evidence":"符号级检索/编辑/重构、find-references、跨文件重构、40+ 语言、记忆系统、多层级 YAML 配置、付费 JetBrains 插件（含免费试用）均可在 material.json summary.evidence 与 verifiedClaims 找到对应 sourceType 依据；integrators 章节列举的 Claude Code/Codex/OpenCode/JetBrains 等与 material 中 topics 一致。"},"signatureVisual":{"passed":true,"evidence":"hero 内含 .constellation 内联 SVG：中心圆心符号 + 放射切线 path（标注 find-references 关系）+ 次级符号 dot + 旋转 orbital 环与 pulse 流动光点，对应 brief.heroGeometry；且显式声明 aria-hidden 仅为关系示意，符合签名视觉存在且真实。"},"mobileTransformation":{"passed":true,"evidence":"max-width:760px 处 hero-grid、grid2 折叠为单列，seccondary 星座图堆叠置顶；每节标题前置 .node 光点标记符号，符合 brief.mobileTransformation 的竖向单列与节前符号标记意图；prefers-reduced-motion 关闭动画。"},"localeParity":{"passed":true,"evidence":"zh-CN.html 与 en.html 结构一致，各 6 节、5 个 h2 + 1 个 h1、相同的 arcband 星座 SVG 与文案对应；qa-report 中 en cjkCount=0，zh-CN 与 en 文案语义等价且自然，技术专名保留原名。"},"antiTemplate":{"passed":true,"evidence":"未回退通用 Hero/终端编辑器壳/手机壳/三张指标卡片/卡片墙；visual 为星座主题，背景仅用径向渐变与极低透明度星场 grain，非发光网格墙；未复用历史种子编辑器面板构图。"},"noFakeUi":{"passed":true,"evidence":"qa-report 两语言桌面/移动 fakeControlCount=0；按钮为指向 github 仓库的真实链接；.sym 等宽行为注释为文本标注，SVG 星座均 aria-hidden，无伪造交互控件。"}},"issues":[]},"repairEvidence":[],"provenance":{"executor":"opencode","model":"deepseek/deepseek-v4-flash","opencodeVersion":"1.1.58","promptVersion":"dodot-showcase-deepseek-v4","inputHash":"a532b16f3b66c4f4579cacc6dc1a7d4945d1a47972005e40ae7e9ba43864284a","artifactHash":"371ed48727386d863dc9ee6adb43dde8c53f9fe9db69dedbb4aa92482116576d","designSkill":"dodot-showcase-design","historicalSeed":{"source":"git_head","sourceCanonicalKey":"github:oraios/serena","assetId":"HEAD:deepseek-v1/oraios-serena","sourceSchemaVersion":"v1","sourceGeneratorVersion":"anthropic:claude-fable-5","sourceArtifactHash":"c1267f67a938c74701230747ec09988d2e16633eb54156ebec0d4f8dcd983207","disposition":"repair_or_regenerate","seedPayloadHash":"bcdbc7f71ff8ce7a27bdbd880e48020b12c50b1f039731dfcc0bdf374fce4875"}},"qaEvidence":{"zh-CN":[{"viewport":"desktop","passed":true,"localePassed":true,"metrics":{"language":"zh-CN","h1Count":1,"h2Count":5,"sectionCount":6,"textLength":1712,"cjkCount":843,"latinLetterCount":553,"scrollWidth":1280,"clientWidth":1280,"scrollHeight":4871,"h1Top":165.796875,"bottomInset":216.703125,"brokenImageCount":0,"missingAltCount":0,"fakeControlCount":0,"emptyLinkCount":0},"screenshotHashes":{"top":"3e943c7aae981be605d8ce18423e43a242b81554021ff11d53dfddcc70b8fd6c","middle":"eccd7b1e0ad51c5057a32c09abd9ca3d8499add134867a4d4573234c7697acaf","bottom":"3f7e4e428812a2aa3aa82ed7c427c2ccf30a6122520ccd70de4dd7c24db537b5"},"consoleErrors":[],"networkRequests":[]},{"viewport":"mobile","passed":true,"localePassed":true,"metrics":{"language":"zh-CN","h1Count":1,"h2Count":5,"sectionCount":6,"textLength":1712,"cjkCount":843,"latinLetterCount":553,"scrollWidth":390,"clientWidth":390,"scrollHeight":5540,"h1Top":159.796875,"bottomInset":216.984375,"brokenImageCount":0,"missingAltCount":0,"fakeControlCount":0,"emptyLinkCount":0},"screenshotHashes":{"top":"af99d2edc3e18c847338299a880b6c62a52b433fd25b02c70cb3ae68a928074a","middle":"2b8c7e9f06d2d4f341201602174d134ae8290ae74ac5b733e9532d10a5060b2d","bottom":"84bd3c1b0a0791e90adc54099f6d9fbc18cfb206b05beb6baa74d252fbf76f48"},"consoleErrors":[],"networkRequests":[]}],"en":[{"viewport":"desktop","passed":true,"localePassed":true,"metrics":{"language":"en","h1Count":1,"h2Count":5,"sectionCount":6,"textLength":3922,"cjkCount":0,"latinLetterCount":3134,"scrollWidth":1280,"clientWidth":1280,"scrollHeight":5097,"h1Top":165.796875,"bottomInset":216.9375,"brokenImageCount":0,"missingAltCount":0,"fakeControlCount":0,"emptyLinkCount":0},"screenshotHashes":{"top":"d0fd88b51c0ad9920ac73a27d5cf6d4bf7263482fba09755d2dfffa9aa3880b8","middle":"78e5885644e364dd53b8c6a276e9367e3c83843519e855cce88231b47ad2dc27","bottom":"164dc0939092db79c3013e095b18d054ba5996409d2eb5dc13d1e50467150c25"},"consoleErrors":[],"networkRequests":[]},{"viewport":"mobile","passed":true,"localePassed":true,"metrics":{"language":"en","h1Count":1,"h2Count":5,"sectionCount":6,"textLength":3922,"cjkCount":0,"latinLetterCount":3134,"scrollWidth":390,"clientWidth":390,"scrollHeight":6072,"h1Top":159.796875,"bottomInset":216.5,"brokenImageCount":0,"missingAltCount":0,"fakeControlCount":0,"emptyLinkCount":0},"screenshotHashes":{"top":"e3396e8bdc8861c67a8f7d8664396532efaf281c1ec4bb619375be2dca1f1fe2","middle":"29abdc13b8ae2e25e54d9cc4df2b217f681dbef5e947e7df66e1800aa82eb621","bottom":"052bf6c2c36417d5323d02f01c0f89a1bb876b8b9accd92a141b25ab48c24330"},"consoleErrors":[],"networkRequests":[]}]}},"generatorVersion":"opencode-1.1.58:deepseek/deepseek-v4-flash:dodot-showcase-deepseek-v4","publishedAt":"2026-08-20T14:21:32.368Z"},"rawExtraction":{"projectRevisionId":"cmt2zh24u01aj6271w7ou795d","projectVersion":1,"sources":[{"canonicalKey":"github:oraios/serena","role":"primary","revisionVersion":1,"url":"https://github.com/oraios/serena"}],"facts":[{"key":"repository_owner","value":"oraios","evidence":[{"path":"repository.owner","excerpt":"oraios"}]},{"key":"repository_name","value":"serena","evidence":[{"path":"repository.name","excerpt":"serena"}]},{"key":"description","value":"A powerful MCP toolkit for coding, providing semantic retrieval and editing capabilities - the IDE for your agent","evidence":[{"path":"repository.description","excerpt":"A powerful MCP toolkit for coding, providing semantic retrieval and editing capabilities - the IDE for your agent"}]},{"key":"primary_language","value":"Python","evidence":[{"path":"repository.language","excerpt":"Python"}]},{"key":"license","value":"MIT","evidence":[{"path":"repository.license","excerpt":"MIT"}]},{"key":"topics","value":"agent, ai, ai-coding, claude, claude-code, codex, ide, jetbrains, language-server, mcp-server, programming, vibe-coding","evidence":[{"path":"repository.topics","excerpt":"agent, ai, ai-coding, claude, claude-code, codex, ide, jetbrains, language-server, mcp-server, programming, vibe-coding"}]},{"key":"source_count","value":"1","evidence":[{"path":"projectRevision.sources","excerpt":"github:oraios/serena"}]}],"taxonomy":{"category":"development_tools","subcategory":"data_ai_tools","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"A powerful MCP toolkit for coding, providing semantic retrieval and editing capabilities - the IDE for your agent"}]},"lifecycle":{"stage":"active","confidence":0.75,"evidence":[{"path":"repository.pushedAt","excerpt":"2026-08-16T14:57:14.000Z"}]},"userIntents":[{"slug":"build_software","label":"构建或改进软件","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"A powerful MCP toolkit for coding, providing semantic retrieval and editing capabilities - the IDE for your agent"}]},{"slug":"automate_work","label":"自动化工作流程","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"A powerful MCP toolkit for coding, providing semantic retrieval and editing capabilities - the IDE for your agent"}]},{"slug":"create_content","label":"创作内容或设计","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"A powerful MCP toolkit for coding, providing semantic retrieval and editing capabilities - the IDE for your agent"}]},{"slug":"analyze_data","label":"分析与理解数据","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"A powerful MCP toolkit for coding, providing semantic retrieval and editing capabilities - the IDE for your agent"}]},{"slug":"learn_explore","label":"学习与探索","confidence":0.65,"evidence":[{"path":"repository.description","excerpt":"A powerful MCP toolkit for coding, providing semantic retrieval and editing capabilities - the IDE for your agent"}]}],"readme":{"headings":["Quick Demo","What Our \"End Users\" Say","How Serena Works","Programming Language Support & Semantic Analysis Capabilities","Language Servers","The Serena JetBrains Plugin","Features","Retrieval","Refactoring","Symbolic Editing","Interactive Debugging","Basic Features","Memory Management","Configurability","Quick Start","User Guide","Acknowledgements"],"commands":[]},"changeSet":[{"field":"authorName","beforeHash":"74234e98afe7498fb5daf1f36ac2d78acc339464f950703b8c019892f982b90b","afterHash":"0f695df6ec72a25fee715b583c8669c378e25a28dc66552fb047e8754b9a4d9e"},{"field":"description","beforeHash":"74234e98afe7498fb5daf1f36ac2d78acc339464f950703b8c019892f982b90b","afterHash":"9071291e3040890015c2b6560bde761f984a066dc9afa228baa25cda28d15aa5"},{"field":"experienceUrl","beforeHash":"74234e98afe7498fb5daf1f36ac2d78acc339464f950703b8c019892f982b90b","afterHash":"d6fb1d6fb62a8300811882522000979856ed2ed690d659950a925c2b8851c679"},{"field":"name","beforeHash":"74234e98afe7498fb5daf1f36ac2d78acc339464f950703b8c019892f982b90b","afterHash":"8bf0689a2db32887eed4ec889e142e9b3aec717b005f5a37276a8ecb12b0040c"},{"field":"repository","beforeHash":"74234e98afe7498fb5daf1f36ac2d78acc339464f950703b8c019892f982b90b","afterHash":"3693916501bbfc6120e75cdf573d8f6496984928db65625f05a0b58b7469004c"},{"field":"sourceCount","beforeHash":"74234e98afe7498fb5daf1f36ac2d78acc339464f950703b8c019892f982b90b","afterHash":"6b86b273ff34fce19d6b804eff5a3f5747ada4eaa22f1d49c01e52ddb7875b4b"}]},"publishedAt":"2026-08-10T10:55:57.621Z","updatedAt":"2026-08-21T13:26:52.705Z","summary":{"id":"cmt2zh24x01al6271epwcfsrw","projectType":"product","oneLine":"为 AI 代理提供语义代码检索与编辑的 MCP 工具包","overview":"Serena 是一个 MCP 工具包，为 AI 编码代理提供类似 IDE 的语义代码检索、编辑、重构和调试能力。它通过符号级理解代码结构，支持超过 40 种编程语言，并可与 Claude Code、Codex 等主流客户端集成。Serena 还提供记忆系统和灵活的 YAML 配置，帮助代理在大型复杂代码库中更高效、可靠地工作。","highlights":["符号级代码检索","跨文件重构支持","40+ 语言支持","记忆系统与配置"],"bestFor":["使用 AI 编码代理的开发者","需要语义代码理解的团队","希望提升代理效率的 Vibe Coding 用户"],"caveats":["JetBrains 插件为付费功能","部分重构能力依赖 JetBrains 插件","评估反馈来自 AI 代理，非人类用户"],"categorySuggestion":"development_tools","subcategorySuggestion":"mcp_integrations","lifecycleStages":["backend_data","testing_quality","deployment_growth"],"audienceIntents":["build","integrate","evaluate"],"confidence":0.9,"evidence":[{"claim":"Serena 是一个 MCP 工具包，为 AI 编码代理提供语义检索和编辑能力，号称“代理的 IDE”。","sourceType":"description","locator":"https://github.com/oraios/serena"},{"claim":"Serena 通过模型上下文协议（MCP）与任何客户端/LLM 集成，支持 Claude Code、Codex、OpenCode、Gemini-CLI 等终端客户端，以及 VSCode、Cursor、JetBrains 等 IDE 插件。","sourceType":"readme","locator":"https://github.com/oraios/serena"},{"claim":"Serena 提供符号级检索、重构、符号编辑和调试工具，支持超过 40 种编程语言（通过语言服务器后端）。","sourceType":"readme","locator":"https://github.com/oraios/serena"},{"claim":"Serena 包含一个记忆系统，用于跨会话、用户和项目共享知识，并可通过 YAML 进行多层级配置。","sourceType":"readme","locator":"https://github.com/oraios/serena"},{"claim":"Serena 提供 JetBrains 插件（付费，有免费试用），利用 JetBrains IDE 的代码分析能力，支持重命名、移动、内联等重构操作。","sourceType":"readme","locator":"https://github.com/oraios/serena"},{"claim":"README 中包含了来自 AI 代理（如 Opus 4.6、GPT 5.4）的评估反馈，称 Serena 显著提升了代码导航和重构效率。","sourceType":"readme","locator":"https://github.com/oraios/serena"}],"publishedAt":"2026-08-17T14:09:07.066Z"},"github":{"url":"https://github.com/oraios/serena","ownerAvatarUrl":"https://avatars.githubusercontent.com/u/181485370?v=4","starsCount":28136,"forksCount":1879,"language":"Python","syncedAt":"2026-08-17T14:06:02.361Z","license":"MIT","defaultBranch":"main","archived":false,"disabled":false,"lastChangedAt":"2026-08-16T14:57:14.000Z","missing":false},"tags":[{"name":"Vibe coding","slug":"vibe-coding","verification":"verified","evidenceUrl":"https://github.com/oraios/serena","evidenceExcerpt":"GitHub topic: vibe-coding"}],"updates":[]},"entryPoint":"ranking","claim":{"state":"unclaimed","verified":false},"initialView":"beautified","community":{"viewer":null,"state":{"likesCount":0,"creativeCount":0,"followersCount":0,"commentsCount":0,"liked":false,"creative":false,"followed":false},"comments":{"comments":[],"nextCursor":null}}}]}],"$L24"]