How real projects brief Claude Code: the commands, conventions and warnings they put in CLAUDE.md.
AdancurusulCLAUDE.mdAdancurusul/serial-mcp-server/CLAUDE.md
locked -- --help cargo run --locked -- serve --help cargo run --locked -- list-ports --json ``` MCP server mode: ```bash cargo run --locked -- serve ``` No-subcommand startup remains compatible with old MCP stdio
HurricaHjzCLAUDE.mdHurricaHjz/second-yourself/CLAUDE.md
# CLAUDE.md — LLM Wiki Schema & Operating Contract > This file is the **schema / governance layer** for
prime-radiant-incCLAUDE.mdprime-radiant-inc/streamlinear/CLAUDE.md
package `@primeradianthq/streamlinear`. It ships two runtime binaries from one TypeScript codebase: - `streamlinear`: the stdio MCP server at `mcp/dist/index.js`. - `streamlinear-cli`: the local CLI at `mcp/dist/cli.js`. The repo also contains `.claude
anombyte93CLAUDE.mdanombyte93/atlas-session-lifecycle/CLAUDE.md
when working with code in this repository. ## Project Overview **Project**: Atlas Session Lifecycle **Goal**: MCP server for AI session lifecycle management — persistent memory, soul purpose tracking, and contract verification
dioptxCLAUDE.mddioptx/mcp-atom-of-thoughts/CLAUDE.md
mcp-atom-of-thoughts (v3) MCP server for structured reasoning via Atom of Thoughts. Decomposes problems into atomic units (premise → reasoning → hypothesis → verification → conclusion) with confidence tracking, session scoping
garagonCLAUDE.mdgaragon/aguara/CLAUDE.md
passing. ## Architecture Aguara is a deterministic static security scanner for AI agent skills and MCP servers. No LLM, no network calls. Go 1.25, module `github.com/garagon/aguara`. ### Public Library API (`aguara.go
merlinhu1CLAUDE.mdmerlinhu1/truthmark/CLAUDE.md
directly. There is no daemon, database, remote service, hidden memory layer, or product-centered MCP server. ## Rules 1. Workflow and skill quality, performance, dispatch clarity, and agent effectiveness
prof-little-bearCLAUDE.mdprof-little-bear/cc-equity-research/CLAUDE.md
mcp.json` already declares the server. - Do **not** fabricate numbers, sketch "what the analysis would look like," or fall back to general knowledge in place of MCP data. If the connector
firishCLAUDE.mdfirish/claude_code_vs/CLAUDE.md
FindTestMethodsAsync`** = Roslyn test discovery). The static-analysis twin of `Debugging/DebuggerReader`; backs the `vs-semantic` MCP server. Roslyn binds **in-proc** (unlike ClrMD) - `VisualStudioWorkspace` is the supported extension entry point. - `src/ClaudeCodeVS/Testing
gregceCLAUDE.mdgregce/tortie/CLAUDE.md
process a script starts, it ends in a `finally` block**, being a shell, a server, a sleeper or a load generator, whatever happened. A script that kills only
BobsilvioCLAUDE.mdBobsilvio/ha-claude/CLAUDE.md
which to call; no keyword-based dispatch - MCP external tools (`mcp.py`) extend the tool set via Model Context Protocol (HTTP or stdio servers) ### Route Structure (`routes/`) 22 Flask blueprints registered
joemckenneyCLAUDE.mdjoemckenney/wake/CLAUDE.md
summarization (Qwen3-0.6B via llama.cpp) - **wake-cli** - CLI binary (`wake`) - **wake-mcp** - MCP server binary (`wake-mcp`) ## Development - `cargo build --release` - Build binaries - `cargo test --all` - Run all tests
timothywarner-orgCLAUDE.mdtimothywarner-org/claude-architect/cca-cert-buddy/CLAUDE.md
five CCA-F exam domains**. It is a real app - Express backend, standalone MCP server, vanilla-TS SPA, SQLite store - not a Copilot agent pack. When you change a layer
adamzhang1987CLAUDE.mdadamzhang1987/kingdee-k3cloud-mcp/CLAUDE.md
Claude Code 工作指南 — kingdee-k3cloud-mcp ## 项目概述 金蝶云星空 MCP Server，基于 FastMCP + kingdee-cdp-webapi-sdk，发布到 PyPI。 ## 每次 commit & push 后的版本检查 完成修改并推送后，**必须主动判断**是否需要升版本号和发布，不要等用户询问。 | 变更类型 | 版本策略 | 是否 Release
fabio-rovaiCLAUDE.mdfabio-rovai/brain-in-the-fish/CLAUDE.md
argument graphs (8 rules) | | `argument_graph` | OWL Turtle → typed argument graph with structural metrics | | `server` | MCP server exposing eval_* and eds_* tools | | `snn` | **Deprecated** — legacy scoring, use `gate` module instead
flupkedeCLAUDE.mdflupkede/codesearch/.claude/CLAUDE.md
Tooling rules (IMPORTANT) - **Use codesearch MCP tools first for discovery** on this repo. The MCP server is verified working and this repo is indexed (alias `codesearch-git`) — `search` / `find` / `explore
stevehuang0115CLAUDE.mdstevehuang0115/crewly/CLAUDE.md
Structure ### Core Components - **Backend** (`/backend/src/`) - Express.js server with TypeScript - **Frontend** (`/frontend/src/`) - React.js with TypeScript - **MCP Server** (`/mcp-server/src/`) - TypeScript MCP server for agent communication - **CLI** (`/cli/src/`) - Command-line interface for Crewly
sysevol-aiCLAUDE.mdsysevol-ai/CodeNib/.claude/CLAUDE.md
ranking, resource guards, skills + runner compiler/ # Two-phase index compilation: IndexCompiler -> RepoManifest mcp/ # Model Context Protocol server (semantic/BM25/regex/Zoekt search) model/ # Retrieval pipelines (bm25/embedding/graph/rerank) + agentless prompts llm/ # litellm-based chat interface
vdaubryCLAUDE.mdvdaubry/bottega/reference/CLAUDE.md
There is no Playwright e2e suite — UI flows are validated manually via the Playwright MCP server and protected by the unit/integration suite (`pnpm test:run`). ## Manual Testing with Playwright MCP
WorkflowsioCLAUDE.mdWorkflowsio/company-os-starter-kit/CLAUDE.md
usage_4}} | ### How Tools Connect (Data Flow) ``` {{source_tool}} -> {{middle_tool}} -> {{destination_tool}} ``` --- ## MCP Server Registry | MCP Server | Type | Purpose | |------------|------|---------| | github | stdio | Repository management, PRs, issues, code search | | {{mcp_server
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.
