@turbyho
基本信息
fw-context builds a compiler-accurate semantic index of your C/C++ firmware from
使用下面的配置,将此服务器添加到你的 MCP 客户端。
{
"mcpServers": {
"fw-context": {
"command": "python3",
"args": [
"-m",
"fw_context_mcp.mcp.server"
]
}
}
}工具列表暂未提供
当前收录暂时没有可展示的工具列表，不代表服务没有工具。请查阅项目文档了解支持的能力。
工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。
Fw Context Mcp is a model context protocol server that builds a compiler-accurate semantic index of embedded C/C++ projects from compile_commands.json and the libclang AST. It exposes tools for LLMs to query symbol definitions, references, call relationships, callback registrations, function pointer assignments, active source content, and other project metadata—giving coding agents the right context instead of raw source files.
Run fw-context init followed by fw-context index --build to create the persistent index, then restart your LLM client. After that, ask questions about your firmware project. Detailed installation and configuration guides are available in the documentation (docs/README.md, docs/installation.md).
compile_commands.jsonLanguage servers are designed for interactive editing (completion, diagnostics, go-to-definition). Fw Context Mcp is built for LLM-assisted reasoning over firmware projects—it exposes high-level semantic queries through MCP, stores a persistent index, and optimizes for repeated questions from an LLM, not for editor latency.
It requires a `compile_commands.json` file from your build system and uses libclang to index the AST. The server is local‑first and works with any LLM client that supports the MCP protocol.
The index is built locally from your project’s compilation database and stored persistently on your machine. No project data is sent to external servers; the only communication is between the MCP server and the LLM client over the local machine.
Fw Context Mcp explicitly requires `compile_commands.json`. It supports Zephyr, PlatformIO, Mbed OS, Arduino, FreeRTOS‑based projects, and custom embedded builds that can produce this file. Without a compilation database, it cannot index the project.
The server does not replace an LLM; it provides better input. It cannot analyze files that are not part of the build index, and its accuracy depends on the quality of the compilation database. It is not designed for interactive editing tasks.
Gives AI agents public URLs (tunnels) for localhost, live HTTP traffic inspection, snapshot publishing, and access control. Part of LocalCan, the ngrok alternative for Mac, Windows and Linux. Free plan.
Read public Instagram profiles and post feeds by handle in Claude, Cursor, and other MCP clients. Get follower counts, captions, hashtags, mentions, and engagement counts as structured JSON.
Console, Backend and CLI to manage Capgo Instant update and Native build for Capacitor apps Build and upload to app store your mobile app
DOM-first AI browser automation for Chrome and MCP clients.
Open-source Node.js FHIR MCP server with SMART Backend Services, metadata-aware search/CRUD tools, compact responses, FHIRPath filtering, safe pagination, audit events, and terminology lookup.
评论