How real projects brief Claude Code: the commands, conventions and warnings they put in CLAUDE.md.
hAcKlycCLAUDE.mdhAcKlyc/MyAgents/CLAUDE.md
小助理 | | `bundled-workspaces/` | 随 App 发布的只读工作区模板源码；复制出的用户工作区不由此目录升级覆盖 | | `specs/` | 当前架构、设计规范、模块技术文档与构建指南 | Sidecar、Plugin Bridge、MCP Server 与 CLI 共用应用内置的单一 Node.js v24，不依赖用户系统安装的 Node。 ## 必须常驻的架构心智模型 ### Owner 与 authority 优先 Owner
reticlehqCLAUDE.mdreticlehq/reticle/CLAUDE.md
that instruments the DOM, network, routing, console, animations, and framework state; a local bridge + MCP server exposes that as structured tools the agent uses to **look, act, observe, and assert
MicrosoftCLAUDE.mdmicrosoft/fabric-toolbox/tools/MicrosoftFabricMgmtMCPServer/CLAUDE.md
CLAUDE.md — MicrosoftFabricMgmt MCP Server Development Guide ## Overview This directory contains a Python-based **Model Context Protocol (MCP) server** that wraps the [MicrosoftFabricMgmt PowerShell module](../MicrosoftFabricMgmt/) and exposes its capabilities
pocketpawCLAUDE.mdpocketpaw/pocketpaw/CLAUDE.md
buffering with edit-in-place (1.5s rate limit). Auto-registers a `pocketpaw-discord` MCP server on startup exposing Discord operations to all MCP-capable backends. Admin commands (`/converse`, `/setstatus
SentryCLAUDE.mdgetsentry/sentry-mcp/packages/mcp-core/src/internal/agents/tools/data/CLAUDE.md
protocols) - **server** - Server-side operations (addresses, ports) - **service** - Service identification (name, version, instance) - **error** - Error information (type, message, stack) - **user** - User identification (id, email, name) - **mcp** - Model Context Protocol attributes
jdevalkCLAUDE.mdjdevalk/specification.website/CLAUDE.md
marketing. For any MDN link or browser-support claim, resolve it through the **MDN MCP server** (`https://mcp.mdn.mozilla.net/`, free, no auth) — it returns the current canonical URL (MDN reorganises
LeslieLeungCLAUDE.mdLeslieLeung/glean/frontend/CLAUDE.md
bash cd frontend/apps/web && pnpm test ``` ## MCP Tools When debugging frontend issues, use `chrome-devtools` MCP to help with: - Taking snapshots and screenshots - Inspecting network requests and console messages - Interacting with
borgheiCLAUDE.mdborghei/Claude-Skills/CLAUDE.md
framework) + memory-tool/context-editing/reasoning-effort/caching upgrades to context-engine, llm-cost-optimizer, agent-workflow-designer, mcp-server-builder. Cross-platform surface unified — one `build_manifest.py` run regenerates `cli/skills.json`, `registry.json`, `.gemini/skills-index.json
borgheiCLAUDE.mdborghei/Claude-Skills/project-management/CLAUDE.md
1on1s/** - 1:1 templates by partner type (EM, designer, IC, manager) **Key Features:** - **Atlassian MCP Server** integration for direct Jira/Confluence operations - **Linear + Notion** integration patterns (GraphQL + REST) - **15+ Python
cyberkaidaCLAUDE.mdcyberkaida/reverse-engineering-assistant/CLAUDE.md
ReVa (Reverse Engineering Assistant) is a Ghidra extension that provides a Model Context Protocol (MCP) server for AI-assisted reverse engineering. It supports both GUI and headless modes, with streamable
cyberkaidaCLAUDE.mdcyberkaida/reverse-engineering-assistant/src/CLAUDE.md
util/CLAUDE.md](main/java/reva/util/CLAUDE.md) - Foundational utilities (AddressUtil, ProgramLookupUtil, etc.) - [plugin/CLAUDE.md](main/java/reva/plugin/CLAUDE.md) - ConfigManager, plugin lifecycle - [server/CLAUDE.md](main/java/reva/server/CLAUDE.md) - MCP server architecture ### Tool Providers - [tools/CLAUDE.md](main/java/reva/tools/CLAUDE.md) - Tool provider patterns and base classes - [tools/decompiler/CLAUDE.md](main/java/reva/tools/decompiler/CLAUDE.md) - Decompiler
golf-mcpCLAUDE.mdgolf-mcp/golf/CLAUDE.md
this repository. ## Project Overview Golf is a Python framework for building MCP (Model Context Protocol) servers with zero boilerplate. It automatically discovers, parses, and compiles Python files containing tools, resources
desplega-aiCLAUDE.mddesplega-ai/agent-swarm/CLAUDE.md
PATH`, use `uvx --from "semble[mcp]" semble` in its place. If the `semble` MCP server is enabled, prefer its `search` / `find_related` tools over the CLI. To keep search output
riba2534CLAUDE.mdriba2534/happyclaw/CLAUDE.md
/agents/{sessionId}/.claude/ ├── ipc/{folder}/ ├── ipc/{folder}/agents/{sessionId}/ ├── memory/ ├── skills/{userId}/ ├── builtin-skills/ ├── mcp-servers/{userId}/ ├── plugins/ ├── agent-profile-runtime/ ├── env/ └── extra/ ``` 主要数据库当前 Schema 版本以 `src/db.ts` 的 `CURRENT_SCHEMA
cyrusagentsCLAUDE.mdcyrusagents/cyrus/CLAUDE.md
session start. This must support subroutine-time updates when allowed/disallowed tools change. For Cursor MCP servers, pre-enable them before session start (`agent mcp list` + `agent mcp enable ` per server
freshtechbroCLAUDE.mdfreshtechbro/claudedesignskills/CLAUDE.md
validate ### Documentation Sources When creating skills, always use the most current documentation: - **Context7 MCP Server** - Primary source for library documentation (see commands above) - **WebFetch** - For official documentation websites when Context7
magicyuan876CLAUDE.mdmagicyuan876/mineru-tianshu/CLAUDE.md
frontend, FastAPI API server, LitServe GPU worker pool, SQLite task queue, optional Redis queue / MCP server / RustFS object storage. Code comments, logs and docs are predominantly Simplified Chinese — match that
VercelCLAUDE.mdvercel/next-devtools-mcp/CLAUDE.md
spawn `dist/index.js` and exercise it over the MCP protocol. Fixtures are in `test/fixtures/`. ## Architecture ### MCP Server Structure The main server entry point is `src/index.ts` which uses the standard MCP
mnemox-aiCLAUDE.mdmnemox-ai/idea-reality-mcp/templates/CLAUDE.md
check proactively. > > Requires: [idea-reality-mcp](https://github.com/mnemox-ai/idea-reality-mcp) configured as an MCP server
H-mmerCLAUDE.mdH-mmer/pentest-agents/CLAUDE.md
distributable product: - Keep Claude Code source assets, non-Claude provider renderers, scaffolded workspace behavior, MCP servers, tests, and docs in sync. - Treat `providers/` as generated output. Do not edit provider
A file Claude Code reads at the start of every session. It holds the commands, conventions and warnings the agent needs for this project.
At the repository root. Claude Code also reads CLAUDE.md files in subdirectories when it works there.
Build and test commands, the project's layout, conventions that aren't obvious from the code, and mistakes to avoid. Short files tend to work better than long ones.
Claude Code reads CLAUDE.md; most other agents read AGENTS.md. Many projects keep one and point the other at it.
