How real projects brief Claude Code: the commands, conventions and warnings they put in CLAUDE.md.
AppleCLAUDE.mdapple/containerization/CLAUDE.md
assembles `bin/containerization-x86_64- .tar.gz` (cctl + cloud-hypervisor + virtiofsd + initfs.ext4 + kernel) for x86_64 Linux deployment, cross-compiled inside the aarch64 dev container via the Static Linux SDK (Swift) and `cargo zigbuild
NVIDIACLAUDE.mdNVIDIA/Isaac-GR00T/CLAUDE.md
generalized humanoid robot skills. The repo contains the model, training pipeline, evaluation harness, and deployment tooling. - **Language:** Python 3.12 - **Package manager:** [uv](https://docs.astral.sh/uv/) - **Build system:** setuptools (see `pyproject.toml
datahaven-xyzCLAUDE.mddatahaven-xyz/datahaven/CLAUDE.md
configurations (mainnet/stagenet/testnet) ├── test/ # E2E testing framework │ ├── suites/ # Test scenarios │ ├── framework/ # Test utilities │ └── launcher/ # Network deployment tools └── deploy/ # Kubernetes deployment charts ``` ### Cross-Component Dependencies - **Contracts → Operator**: AVS contracts register/slash operators
WenyuChiouCLAUDE.mdWenyuChiou/awesome-agentic-ai-zh/CLAUDE.md
current function-calling exercises | | **`qwen3.5:4b`** | Stage 7 (debate / eval / observability / streaming / deploy mechanics) | 3.4 GB official Ollama tag. These exercises do not depend on function calling; this row does
czlonkowskiCLAUDE.mdczlonkowski/n8n-skills/CLAUDE.md
expert guidance on using n8n-mcp MCP tools effectively for building n8n workflows, plus deploying the self-hosted n8n that runs them. **Architecture**: - **n8n-mcp MCP Server**: Provides data access
fengshao1227CLAUDE.mdfengshao1227/ccg-workflow/dsh-ccg/CLAUDE.md
only through the shell's own static registry, which is a property of the deployment, not a contract. Confirmed live: the dock's dot renders with
CloudflareCLAUDE.mdcloudflare/vibesdk/CLAUDE.md
bundling, and App Facets - `/sdk` - TypeScript client SDK - `/migrations` - D1 migrations - `/scripts` - Setup and deployment utilities ## Key Architectural Patterns **ThinkAgent:** - One Agent backed by a Durable Object per app session
weave-osCLAUDE.mdweave-os/router/CLAUDE.md
Reserve `panic` for startup-time fail-fast (`config.MustGet`, cluster-scorer boot failure, invalid `ROUTER_DEPLOYMENT_MODE`) where misconfiguration must abort process. - **Never introduce DI container, reflection-based wiring, or service
weave-osCLAUDE.mdweave-os/router/cmd/CLAUDE.md
boot, `main.go` panics. Misconfiguration must abort the process rather than silently degrade. - Validate deployment target overrides through `policy.Registry.ValidateDeployment` after provider/model configuration is resolved. A partial or catalog-incompatible explicit target
VercelCLAUDE.mdvercel/examples/starter/cms-sanity-graphql-fragments/CLAUDE.md
format` (Biome formatter) - **Sanity Studio**: `npm run sanity:dev` (start Sanity Studio development server) - **Deploy GraphQL**: `npm run sanity:deploy` (deploy GraphQL API to Sanity) - **Generate Types**: `npm run schema
PeonPingCLAUDE.mdPeonPing/peon-ping/CLAUDE.md
feature relates to CESP/MCP/packs) ## Website `docs/` contains the static landing page ([peonping.com](https://peonping.com)), deployed via Vercel. A `vercel.json` in `docs/` provides the `/install` redirect so `curl -fsSL peonping.com/install
MicrosoftCLAUDE.mdmicrosoft/poml/CLAUDE.md
extension: - `poml.languageModel.provider`: LLM provider (openai, openaiResponse, microsoft, anthropic, google) - `poml.languageModel.model`: Model name or deployment - `poml.languageModel.apiKey`: API authentication key - `poml.languageModel.apiUrl`: Custom API endpoint - `poml.languageModel.temperature`: Sampling temperature (0-2) ## Testing Strategy - **Unit
PydanticCLAUDE.mdpydantic/logfire/CLAUDE.md
particularly to check files that haven't been changed. # Documentation Docs are rendered and deployed through the `pydantic/unified-docs` pipeline. Do not use MkDocs checks in this repository. When linking between
macro-incCLAUDE.mdmacro-inc/macro/CLAUDE.md
patterns, and browser verification. | | `apps/`, `packages/` | Other apps and shared frontend packages. | | `services/` | Deployable services and workers; Rust package names come from each `Cargo.toml`. | | `crates/` | Reusable backend libraries, domain services
txn2CLAUDE.mdtxn2/kubefwd/CLAUDE.md
forwarding lifecycle events ### Demo Environment A comprehensive demo manifest is available for testing: ```bash # Deploy 60 services across 2 namespaces kubectl apply -f test/manifests/demo-microservices.yaml # Forward all demo services with
gadievronCLAUDE.mdgadievron/raptor/CLAUDE.md
# RAPTOR - Autonomous Offensive/Defensive Research Framework Safe operations (install, scan, read, generate): DO IT. Dangerous
samberCLAUDE.mdsamber/cc-skills-golang/CLAUDE.md
# CLAUDE.md ## Project Overview This is a Claude Code plugin containing AI agent skills for
AnthropicCLAUDE.mdanthropics/commerce-agents/CLAUDE.md
/marketplace.json` points at it. - `docs/`: `safety.md`, `backends.md`, `deployment.md`. `scripts/`: install, demo, smoke, screenshots, check, deploy, verify. - `tests/`: the suites that span packages (both roles on all three paths); each package
aaron-he-zhuCLAUDE.mdaaron-he-zhu/aaron-marketing-skills/CLAUDE.md
kernel. **Lean** uses a generated Skill capsule plus that kernel. Both are unavailable for deployment until paired evidence satisfies [`references/prompt-profiles.json`](references/prompt-profiles.json). - Evaluation-only compact assemblies are visibly `deployment_eligible: false
datopianCLAUDE.mddatopian/portaljs/CLAUDE.md
push to R2, and write a source-vs-derived parity report - `/portaljs-deploy` — one-shot deploy to Vercel or static hosting - `/portaljs-connect-ckan` — wire a portal to a CKAN backend over
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.
