What it does
Fast polyglot source code intelligence: symbols, imports, dependencies, and impact analysis.
- Connects 在你的电脑上运行
- Account or key No key needed
- Category 开发者工具
可搭配使用
教 AI 做这类工作的 Skill,可与本项一起使用
andrej-karpathy-skills
A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.
graphify
Turn any codebase, with its docs, SQL schemas, configs, and PDFs, into a queryable knowledge graph. A /graphify skill for Claude Code, Cursor, Codex, and Gemini CLI: local deterministic AST parsing, every edge explained, no vector store.
为什么排在这里
列表按使用分(满分 100)排序,由以下部分组成
| 热度 | 6/80 | GitHub 星标或每周下载量:较大的计 60,较小的计 20 |
|---|---|---|
| 新鲜度 | 7/12 | 30 天内有发布或代码更新得满分 |
| 仍在响应 | 0/8 | 上次检查时云端端点有响应 |
| 使用分 | 13/100 |
质量: 基础检查通过 3/3 项(仅展示,不用于排序) · 上升势头: 暂无数据 (只用于“上升中”标签)
安装
选择你使用的 AI 应用,然后复制设置。
通过 3/3 项检查 · 在你的电脑上运行
第一步:你用哪个 AI?
选择你用的应用,步骤会显示在下方。
可连接 Claude,但需要先准备
不需要密钥,不用安装程序 · 仅限电脑上的 Claude 应用(Windows 或 Mac),网页和手机不可用
双击文件,然后在电脑版 Claude 应用中点击 Install
试一试
用 Symbols 告诉我你能帮我做什么,然后做一件给我看✓ 如果 Claude 说正在使用 Symbols,或回答中出现 Symbols 的工具名称,就表示连接成功
菜单名称已于 2026年11月11日 对照 Anthropic 帮助页面核实 · 你的界面可能因版本略有不同
ChatGPT 暂时无法连接 Symbols
Symbols 需要安装在你的电脑上,而 ChatGPT 只能连接在网上运行的服务器
你可以改用这些方式
A · 使用能连接 Symbols 的应用
- 不需要密钥,不用安装程序 · 仅限电脑上的 Claude 应用(Windows 或 Mac),网页和手机不可用
B · 开发者工具 类中能连接 ChatGPT 的其他服务器
从同类中使用较多的里挑选,仅限用地址即可连接、无需密钥的
- Sentry✓ ChatGPT
- Compiler Explorer✓ ChatGPT
- GitDiagram✓ ChatGPT
Claude Code 暂时无法连接 Symbols
开发者只提供了电脑版 Claude 应用的安装文件
你可以改用这些方式
A · 使用能连接 Symbols 的应用
- 不需要密钥,不用安装程序 · 仅限电脑上的 Claude 应用(Windows 或 Mac),网页和手机不可用
Cursor 暂时无法连接 Symbols
开发者只提供了电脑版 Claude 应用的安装文件
你可以改用这些方式
A · 使用能连接 Symbols 的应用
- 不需要密钥,不用安装程序 · 仅限电脑上的 Claude 应用(Windows 或 Mac),网页和手机不可用
VS Code 暂时无法连接 Symbols
开发者只提供了电脑版 Claude 应用的安装文件
你可以改用这些方式
A · 使用能连接 Symbols 的应用
- 不需要密钥,不用安装程序 · 仅限电脑上的 Claude 应用(Windows 或 Mac),网页和手机不可用
一览
哪些应用可以使用它
依据这个服务器的连接方式和各应用自己的文档,检查于 2026年11月10日。
在你的电脑上运行,适用于
ClaudeGooseLM StudioMsty StudioCherry StudioVisual StudioWindsurf (Devin Desktop)ZedKiroClineContinueJetBrains AI AssistantCodex CLIGemini CLIOpenCodenodecodeWarpLibreChatRaycast
它能做什么
它在你的电脑上运行;安装后,工具列表会显示在你的 AI 应用中。
基础检查i
我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月19日
- ✓ 通过
作者与仓库一致
注册名称已通过 GitHub 账户 Jordan-Horner 验证,该账户拥有此仓库
- ✓ 通过
仍在维护
代码最后更新于3个月前
- ✓ 通过
有许可证
使用 MIT
- – 不适用
端点有响应
它在你的电脑上运行,没有互联网端点
- – 不适用
使用 OAuth 登录
在你的电脑上运行,改用本地设置
- … 尚未检查
工具描述中没有隐藏指令
它在你的电脑上运行;我们只能读取互联网端点的工具列表
作者:申请重新检查或报告信息错误
给作者:在 README 中加徽章
如果你是作者,可把这段代码贴到 README。徽章会链接回本页。
[](https://mcp.popnx.com/en/servers/jordan-horner-symbols)讨论
提问、分享技巧或报告无法使用。按“有帮助”数量排序。
用户评价
还没有评价
与“开发者工具”中的其他服务器比较i
- 星标数多于此分类中其他 6116 个服务器的 77%
- 发布到注册中心的时间比此分类中其他 6116 个服务器的 5% 更新
- 基础检查结果等于或优于此分类中其他 6116 个服务器的 99%
与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。
常见问题
Symbols 安全吗?
它在 3 项适用的基础检查中通过了 3 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。
它需要哪些密钥或代码?
注册中心没有列出必需的密钥,不过有些服务器会在首次使用时要求你登录。
它在我的电脑上运行还是在云端?
它在你的电脑上运行。你的 AI 应用会在需要时启动它,数据不会经过作者的服务器(除非它自己调用在线服务)。
它还在维护吗?
GitHub 仓库最后更新于3个月前,最新版本(1.0.3)于8个月前进入注册中心。
哪些应用可以使用它?
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。你的电脑上还需要有 MCPB。
有什么可以替代它?
同一分类中有:GitDiagram, Appllama, Cloudflare MCP, Omnara, Repowise
功能相近的其他选择
同一分类、其他作者、命令已通过检查。官方的排在前面,再按星标排序。
如何看卡片:↓ 每周下载量i★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i
GitDiagram
Architecture diagrams, explanations and Mermaid source for public GitHub repositories.
Appllama
Study screens, flows and paywalls from top-earning iOS apps, then build from what wins.
Omnara
Deploy durable AI agents with Omnara, an open-source managed-agent platform.
Repowise
Codebase intelligence for AI coding agents: graph, git history, docs, decisions, code health.
| 本项 Symbols | GitDiagram | Appllama | Cloudflare MCP | Omnara | Repowise | |
|---|---|---|---|---|---|---|
| 运行位置 | 需要安装在电脑上 | 在云端运行,无需安装 | 在云端运行,无需安装 | 在云端运行,无需安装 | 在云端运行,无需安装 | 云端或安装在电脑上都可以 |
| 需要密钥 | 否 | 否 | 否 | 否 | 否 | 否 |
| 可在 Claude 中按 + 添加 | 否 | 否 | 否 | 是 | 否 | 否 |
| 基础检查 | 3/3 | 4/4 | 4/4 | 4/4 | 4/4 | 5/5 |
| 许可证 | MIT | MIT | MIT | Apache-2.0 | Apache-2.0 | AGPL-3.0 |
| 价格 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 |
作者的 README
打开 README
按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。
symbols
A fast, polyglot source code intelligence CLI. Extract symbols, parse imports, trace dependencies, and analyze impact — all from the command line.
No language server required. No build step for your projects. Just point it at your code.
Table of contents
- Why this exists
- How it saves context
- What it does
- Install
- Language support
- Usage
- Symbol extraction
- Import parsing
- Dependency queries
- Impact analysis
- Project graph summary
- JSON output
- Shorthand
- Symbol search
- MCP server
- How it works
- Project root detection
- Limitations
- License
Why this exists
symbols is for the moments when you need to understand a codebase quickly without opening 30 files first.
Common pain points it targets:
- You are about to change a file and need to know blast radius immediately.
- You are onboarding to an unfamiliar repo and need a map, not a scavenger hunt.
- You are reviewing a PR and want concrete dependency and ownership signals.
- You are using AI coding tools and need reliable, structured project context on demand.
Instead of manually reconstructing context from editor tabs, grep output, and memory, symbols gives you the structural view in one step.
How it saves context
symbols saves context in two practical ways:
- It externalizes code structure into fast, repeatable queries (
list,deps,dependents,impact,graph,search) so you do not have to rebuild mental maps every session. - It exposes the same model through MCP (
syms mcp) so agents and tools can fetch fresh project facts directly, rather than relying on stale chat history or guessed file relationships.
Net effect:
- less re-reading
- fewer "what will this break?" surprises
- faster onboarding and safer refactors
- more useful AI assistance because context is retrieved, not improvised
- lower cost from fewer exploratory engineering cycles and reduced AI token spend on repo re-discovery
What it does
syms list server.py # functions, classes, constants, variables
syms imports server.py # parsed import statements syms deps server.py # files this file imports from syms dependents server.py # files that import this file syms impact server.py # full impact analysis (direct + transitive) syms graph . # project-wide dependency summary syms search User # find symbols by name across a project syms mcp # run as MCP server for AI tools
Install
Option 1: Build from source
git clone https://github.com/Jordan-Horner/symbols.git
cd symbols go build -o syms . sudo mv syms /usr/local/bin/
Requirements: Go 1.26+
Option 2: Direct installation (Linux/macOS)
# Install directly to /usr/local/bin
curl -L https://github.com/Jordan-Horner/symbols/releases/latest/download/syms-$(uname -s)-$(uname -m) -o /usr/local/bin/syms chmod +x /usr/local/bin/syms
Option 3: Homebrew (macOS)
brew tap Jordan-Horner/tap
brew install syms
Verify installation
syms --version
Language support
Symbol extraction uses tree-sitter for full AST parsing (function signatures with parameters, classes, types, constants). Import parsing and dependency resolution use regex.
| Language | Symbols | Import parsing | Dependency resolution |
|---|---|---|---|
| Python | tree-sitter (functions, classes, constants, variables) | regex | Relative + absolute imports |
| TypeScript | tree-sitter | regex | tsconfig.json path aliases, relative paths, index.ts |
| JavaScript | tree-sitter | regex | Same as TypeScript (also reads jsconfig.json) |
| Svelte | tree-sitter (script block) | regex | Same as TypeScript |
| Go | tree-sitter | regex | go.mod module prefix, package directories |
| Java | tree-sitter | regex | Dot-to-slash, src/main/java prefix |
| Kotlin | tree-sitter | regex | Same as Java + .kt |
| Rust | tree-sitter | regex | crate/self/super, mod.rs |
| C# | tree-sitter | regex | Namespace-to-path, class name fallback |
| PHP | tree-sitter | regex | PSR-4 conventions, require/include |
| C/C++ | tree-sitter | — | — |
| Ruby | tree-sitter | — | — |
| Scala | tree-sitter | — | — |
| Bash | tree-sitter | — | — |
Usage
Symbol extraction
# Single file
syms list app.py
Multiple files
syms list src/main.go src/handlers.go
Recursive directory scan
syms list -r src/
JSON output (for piping to other tools)
syms list --json app.py
Pretty JSON output (human-readable)
syms list --json --pretty app.py
Optional: include precise symbol ranges
syms list --json --ranges app.py
Count symbols per file
syms list --count src/
Filter by symbol kind (repeatable or comma-separated)
syms list --filter class src/ syms list --filter class,function src/ syms list --filter class --filter function src/
Output:
### `app.py` — 245 lines
constant VERSION # line 1 constant API_URL # line 3 variable app # line 5 class Application # line 12 def init(self, config) # line 15 async def start(self) # line 34 def shutdown(self) # line 78
Import parsing
syms imports server.py
Output:
### `server.py`
from flask import Flask, jsonify # line 1 from .models import User, Post # line 2 import os # line 3
Dependency queries
# Direct dependencies
syms deps src/handlers.go
Transitive (everything it depends on, recursively)
syms deps -t src/handlers.go
Who imports this file?
syms dependents src/models.py
Transitive dependents
syms dependents -t src/models.py
Impact analysis
syms impact src/core/utils.py
Output:
### `src/core/utils.py` — impact analysis
Direct dependents: 8 Transitive dependents: 23
Direct: src/api/handlers.py src/core/auth.py src/core/db.py ...
Indirect (transitive): src/api/routes.py src/main.py tests/test_auth.py ...
Project graph summary
syms graph .
Output:
Project dependency graph
Files: 187 Import edges: 562 Unresolved imports: 43
Most depended-on files: src/utils.py (36 dependents) src/config.py (33 dependents) src/models.py (23 dependents)
Heaviest importers: src/app.py (28 imports) src/main.py (24 imports)
Circular dependencies (1): src/config.py <-> src/runner.py
JSON output
All commands support --json for machine-readable output:
syms impact --json src/utils.py | jq '.direct_dependents'
syms graph --json . | jq '.hot_spots[:5]'
Optional: pretty-print JSON for humans
syms graph --json --pretty .
Full edge map (file → its dependencies)
syms graph --json . | jq '.edges'
What does a specific file depend on?
syms graph --json . | jq '.edges["src/app.py"]'
Shorthand
The list subcommand is the default — you can omit it:
# These are equivalent:
syms list app.py syms app.py
Flags work too:
syms -r src/ --json
Symbol search
# Find symbols by name (fuzzy: exact > prefix > contains)
syms search User
JSON output
syms search --json handle
Search in a specific project
syms search --root /path/to/project Config
Search only specific symbol kinds
syms search --filter class User
Optional: include precise symbol ranges in search results
syms search --json --ranges User
Output:
Found 3 symbols matching "User":
class User models.py:1 class UserProfile models.py:5 function get_user(id) api/handlers.py:12
MCP server
Run syms as an MCP server for AI tool integration (e.g. Claude Code):
syms mcp
Exposes all functionality as MCP tools over stdio (JSON-RPC 2.0):
| Tool | Description |
|---|---|
syms_list | Extract symbols from files |
syms_imports | Parse import statements |
syms_deps | File dependencies |
syms_dependents | Reverse dependencies |
syms_impact | Impact analysis |
syms_search | Search symbols by name |
syms_graph | Project dependency graph |
syms_list and syms_search accept optional kinds: string[] arguments to filter symbol kinds. syms_list and syms_search also accept optional include_ranges: boolean for start/end line+column metadata. Tool results are returned in structuredContent (not JSON text blobs in content[].text).
Claude Code setup
After installing syms, configure it as an MCP server:
Project-level (recommended for teams):
Create .mcp.json in your project root:
{
"mcpServers": { "symbols": { "command": "syms", "args": ["mcp"] } } }
Commit this file so your team gets the symbols server automatically.
Global (all projects):
Create or edit ~/.mcp.json:
{
"mcpServers": { "symbols": { "command": "syms", "args": ["mcp"] } } }
After configuration:
- Restart Claude Code
- When prompted, approve the
symbolsMCP server - Claude Code will now have access to code intelligence tools in all your projects
How it works
Symbol extraction uses tree-sitter for full AST parsing. Each language has a compiled grammar (linked statically into the binary) that produces a syntax tree. The tool walks the tree to extract top-level declarations with names, kinds, line numbers, and function parameters. For Python, module-level assignments are also extracted as constants (UPPER_CASE) or variables.
Import parsing uses regex patterns tuned to each language's import syntax. This is fast and reliable for standard import forms without needing AST parsing.
Dependency resolution maps import specifiers to actual files on disk using language-specific conventions:
- Python: module dot-path to file path, relative import resolution
- Go:
go.modmodule name stripping, package-to-directory mapping - Java/Kotlin: dot-to-slash convention, standard source root prefixes (
src/main/java/) - Rust:
crate/self/superpath resolution,mod.rsconvention - C#: namespace-to-path with progressive prefix stripping
- PHP: PSR-4 backslash-to-slash mapping,
require/includepath resolution
Directory scanning uses early pruning of .git, node_modules, dist, build, vendor, target, and other common non-source directories.
Project root detection
For deps, dependents, impact, and graph, the tool auto-detects the project root by walking up the directory tree looking for .git, package.json, or pyproject.toml. Override with --root:
syms deps src/app.py --root /path/to/project
Limitations
- Convention-based resolution — dependency resolution uses file path conventions, not compiler/build system integration. TypeScript/JavaScript
pathsfromtsconfig.json/jsconfig.jsonare supported (includingextends), but webpack/vite aliases defined outside tsconfig are not. - File-level granularity — dependencies are traced at the file level (import graph), not at the function or symbol level. There is no call graph.
- C/C++ includes — `#include