菜单
RustFS — S3-compatible distributed object storage written in Rust. Use when building with RustFS or asking about its APIs, configuration, deployment, Docker setup, or integration. Fetch live documentation for up-to-date details.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Principles and a linter for writing code that AI coding agents can read and modify correctly. Use when writing, refactoring, reviewing, or structuring code that will be maintained partly by AI agents (Claude Code, Cursor, Copilot, Aider, Devin). Also use when the user asks about agent-friendly code, AI-friendly architecture, how to structure a repo for AI maintenance, why agents fail on certain codebases, whether SOLID/DRY still apply with AI, or wants to audit an existing codebase for AI-readability. Applies proactively when making naming, file-organization, abstraction, or dependency choices in code the user is likely to maintain with agents — even if they don't explicitly ask for agent-friendly output. Ships with a zero-dependency linter (scripts/lint.py) that flags concrete anti-patterns.
Better Auth — framework-agnostic authentication and authorization framework for TypeScript. Use when building with Better Auth or asking about its APIs, configuration, plugins, session management, OAuth, or integration. Fetch live documentation for up-to-date details.
BullMQ — Redis-based distributed job and message queue for Node.js with workers, schedulers, flows, and rate limiting. Use when building with BullMQ or asking about its queues, workers, job configuration, repeatable jobs, flow producers, or Redis connection setup. Fetch live documentation for up-to-date details.
Bun — fast all-in-one JavaScript/TypeScript runtime, package manager, bundler, and test runner. Use when building with Bun, running TypeScript, managing packages with bun install, writing tests with bun test, or asking about Bun APIs, configuration, or Node.js migration. Fetch live documentation for up-to-date API details.
Drizzle ORM — lightweight, TypeScript-first SQL ORM with schema-as-code and zero-dependency query builder. Use when building with Drizzle ORM or asking about its schema definition, migrations, query builder, relations, or integration with databases like PostgreSQL, MySQL, or SQLite. Fetch live documentation for up-to-date details.
Elysia — Bun-native web framework with end-to-end type safety and Eden Treaty typed client. Use when building with Elysia or asking about its routing, plugins, lifecycle hooks, validation, Eden Treaty, WebSocket, SSE, or integration with Bun. Fetch live documentation for up-to-date details.
| name | rustfs |
| description | RustFS — S3-compatible distributed object storage written in Rust. Use when building with RustFS or asking about its APIs, configuration, deployment, Docker setup, or integration. Fetch live documentation for up-to-date details. |
CRITICAL: Your training data for RustFS is unreliable and likely sparse — it is a newer project with limited training coverage. Before writing any code, you MUST use
WebFetchto read the current README:
WebFetch("https://raw.githubusercontent.com/rustfs/rustfs/main/README.md")Do not proceed without fetching this URL first. Never assume an API or config option exists — verify against current docs.
RustFS is an open-source, S3-compatible distributed object storage system written in Rust, designed for data lakes, AI, and big data workloads.
Set Docker volume ownership to UID 10001. The container runs as non-root user rustfs (UID 10001). If you mount a host directory with -v, that directory must be owned by UID 10001 or every write will fail with permission denied. Run chown -R 10001:10011 /your/data/path before starting the container.
Default credentials are rustfsadmin / rustfsadmin, not MinIO defaults. The S3 API listens on port 9000; the management console is on port 9001. Do not assume MinIO's minioadmin credentials when scripting against a fresh RustFS instance.
Do not use distributed mode in production yet. As of early 2026, distributed mode is still under testing. Single-node mode is the only production-ready deployment target. Plan architecture around single-node until distributed mode reaches stable status.
