{
  "markdown": "# DoiT MCP Skills\n\nAI agent skills that bring [DoiT Cloud Intelligence](https://www.doit.com/) into your favorite AI clients — Claude Code, ChatGPT, Cursor, Codex, and Gemini CLI.\n\nQuery cloud costs, investigate anomalies, generate reports, and produce API integration code across AWS, GCP, Azure, Snowflake, Datadog, and more — all through natural language.\n\n## Skills\n\n\n| Skill                                                                        | Description                                                                              |\n| ---------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------- |\n| **[doit-mcp-setup](skills/doit-mcp-setup/)**                                 | Configure DoiT MCP in any supported AI client with correct credentials and OAuth         |\n| **[doit-mcp-reporting](skills/doit-mcp-reporting/)**                         | Generate cost intelligence reports and run ad-hoc Cloud Analytics queries                |\n| **[doit-mcp-anomaly-investigation](skills/doit-mcp-anomaly-investigation/)** | Root-cause cost anomalies by correlating DoiT alerts with cloud-specific inspection      |\n| **[doit-mcp-api](skills/doit-mcp-api/)**                                     | Generate production-ready API call code (TypeScript, Python, Go, Bash) for DoiT Platform |\n\n\n## Quick Start\n\n### Claude Code (Plugin)\n\nInstall as a plugin for the full experience — skills, MCP server, and tools in one package:\n\n```bash\nclaude plugin marketplace add https://github.com/doitintl/doit-mcp-skills\nclaude plugin install doit-mcp@doit-mcp-skills\n```\n\nThen optionally run the setup skill to configure credentials:\n\n```\n/doit-mcp-setup\n```\n\n### Claude Code (Skills Only)\n\nIf you only want the skills without the plugin wrapper:\n\n```bash\nclaude install-skill https://github.com/doitintl/doit-mcp-skills\n```\n\nThen optionally run the setup skill to configure credentials:\n\n```\n/doit-mcp-setup\n```\n\n### Other AI Clients (Cursor, Codex, Gemini CLI, ChatGPT)\n\nThe skills in this repo are plain Markdown files — any AI that can read a file can use them. To set up DoiT MCP in a non-Claude client:\n\n1. **Point your agent at the skill file** — paste the contents of the relevant `SKILL.md` into your agent's context, or reference the file directly if your client supports file attachments.\n2. **Run setup** — tell your agent: *\"Follow the instructions in this skill to configure DoiT MCP for [your client].\"*\n3. The agent will write the correct config file and guide you through any OAuth steps.\n\nAlternatively, ask your agent to run the **doit-mcp-setup** skill for your client:\n\n- **Cursor** — per-project `.cursor/mcp.json` configuration\n- **ChatGPT** — manual setup via Actions UI (requires developer mode + API key)\n- **Codex** — `~/.codex/config.toml` configuration\n- **Gemini CLI** — `~/.gemini/settings.json` configuration\n\n### DCI CLI (Optional)\n\nThe setup skill can also install the **DoiT Cloud Intelligence CLI** — a terminal companion that talks to the same DoiT API as the MCP server. Run the setup skill and choose the CLI option, or install it directly:\n\n| Platform | Command |\n| -------- | ------- |\n| macOS | `brew install doitintl/dci-cli/dci` |\n| Windows (WinGet) | `winget install DoiT.dci` |\n| Windows (Scoop) | `scoop bucket add doitintl https://github.com/doitintl/dci-cli && scoop install dci` |\n| Linux (.deb / .rpm) | Download from [Releases](https://github.com/doitintl/dci-cli/releases/latest) |\n\nAfter install, run `dci login` to authenticate and `dci list-reports` to verify.\n\n## What You Can Do\n\n**Cost Reporting**\n\n> \"Show me my top 5 cost drivers for the last 30 days\"\n> \"Break down GCP costs by project for Q1\"\n\n**Anomaly Investigation**\n\n> \"Show me recent cost anomalies\"\n> \"Investigate the EC2 cost spike from last Tuesday\"\n\n**API Code Generation**\n\n> \"Generate TypeScript code to query costs by service for the last month\"\n> \"Show me a Python script that lists all my cloud reports\"\n\n## Architecture\n\n```\ndoit-mcp-skills/\n├── .claude-plugin/          # Claude Code plugin metadata\n│   ├── plugin.json          # Plugin manifest (skills + MCP config)\n│   └── marketplace.json     # Marketplace manifest for plugin install\n├── .mcp.json                # MCP server configuration\n└── skills/\n    ├── doit-mcp-setup/      # Client setup & configuration\n    ├── doit-mcp-reporting/  # Report generation & queries\n    ├── doit-mcp-anomaly-investigation/\n    └── doit-mcp-api/        # API code generation\n```\n\nEach skill contains:\n\n- `SKILL.md` — skill instructions and workflow\n- `agents/openai.yaml` — agent configuration and tool dependencies\n- `references/` — detailed reference docs (dimensions, metrics, field mappings)\n- `templates/` — config snippets or output templates\n- `examples/` — working code examples (where applicable)\n\n## How It Works\n\nAll skills connect to the DoiT MCP server at `https://mcp.doit.com/sse` — no local server installation required. Authentication is handled via OAuth (automated clients) or API key (manual clients like ChatGPT).\n\n**Supported platforms**: AWS, GCP, Azure, Snowflake, Datadog, Databricks, OpenAI\n\n## Prerequisites\n\n- A [DoiT](https://www.doit.com/) account with **Billing Profiles Admin** permission\n- An AI client that supports MCP (Claude Code, Cursor, ChatGPT, Codex, or Gemini CLI)\n\n## License\n\n[MIT](LICENSE)",
  "bytes": 5409,
  "sha": "a28e2a077a8d819105226ccc6dfe83dc09baf9d896b9cd3e3ac396dbd02c3a6c",
  "repo_slug": "doitintl/doit-mcp-skills",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_doitintl_doit_mcp_skills_doit_mcp_skills_ad8149b7/readme"
}