{
  "markdown": "# datahub-skills\n\nAgent skills for working with DataHub — plan and review connectors, search the catalog, enrich metadata, trace lineage, manage data quality, and set up connections. Works with [Claude Code](https://claude.ai/claude-code), [Cortex Code](https://docs.snowflake.com/en/user-guide/cortex-code/cortex-code), [Cursor](https://cursor.sh), [Codex](https://openai.com/codex), [Copilot](https://github.com/features/copilot), [Gemini CLI](https://github.com/google-gemini/gemini-cli), [Windsurf](https://windsurf.com), and other [Agent Skills](https://skills.sh)-compatible tools.\n\n## What's in here\n\n### Catalog interaction skills\n\n#### Search\n\nSearch the DataHub catalog, discover entities, and answer ad-hoc questions about your data. Supports keyword search, filtered browse, column-name search, structured property queries, and multi-step question answering.\n\n```\n> Find revenue tables in Snowflake\n> Who owns the customer pipeline?\n> /datahub-search datasets tagged PII\n```\n\n#### Enrich\n\nAdd or update metadata in DataHub — descriptions, tags, glossary terms, ownership, and deprecation. Shows a before/after plan and asks for approval before making changes.\n\n```\n> Add a description to the orders table\n> Tag these columns as PII\n> /datahub-enrich set owner of revenue_daily to @jdoe\n```\n\n#### Lineage\n\nExplore data lineage, trace upstream sources and downstream consumers, perform impact analysis, and map cross-platform data flows.\n\n```\n> What feeds into the revenue dashboard?\n> Impact analysis for changing the orders table\n> /datahub-lineage trace the customer pipeline\n```\n\n#### Quality\n\nManage data quality — create and run assertions (freshness, volume, SQL, field, schema), set up smart AI-inferred assertions, raise and resolve incidents, and configure notification subscriptions. Separates Open Source (diagnostic) from Cloud (full management) capabilities.\n\n```\n> Find datasets with failing assertions\n> Create a freshness assertion on the orders table\n> /datahub-quality raise an incident on the customer pipeline\n> Subscribe me to assertion failures via Slack\n```\n\n#### Setup\n\nInstall the DataHub CLI, configure authentication, verify connectivity, and set up default scopes and profiles for the other interaction skills.\n\n```\n> Set up my DataHub connection\n> /datahub-setup focus on Snowflake in the Finance domain\n> Create a profile for the data-eng team\n```\n\n#### Evals\n\nRuns your saved DataHub evals and reports how they did. An eval is a saved question plus the conditions its answer must meet.\n\nEach eval is answered in a fresh agent with the DataHub tools attached, then reported back **without a verdict** so DataHub scores it with the same judge it uses for its own runs — which is what makes two runs comparable. The same command reports an answer produced anywhere else, so you can score another agent's answer against the same conditions.\n\n```\n> Run our evals\n> /datahub-evals run urn:li:eval:abc123\n> /datahub-evals score this Slack answer against that eval\n```\n\nNeeds the DataHub Cloud CLI — `pip install 'acryl-datahub-cloud[datahub-evals]'` — and a DataHub MCP server for the answering agent.\n\n### Connector development skills\n\n#### Connector planning\n\nWalks you through building a new DataHub connector in four steps: classify the source system type, research it (using a dedicated agent or inline), generate a `_PLANNING.md` with entity mapping and architecture, and get your sign-off before anyone writes code.\n\n```\n> Plan a connector for ClickHouse\n> /connector-planning duckdb\n```\n\n#### Connector review\n\nChecks connector code against the 22 standards (see below). On Claude Code it runs five agents in parallel — silent failures, test coverage, type design, simplification, comment resolution. On other platforms it does the same checks one at a time.\n\n```\n> Review my connector\n> /connector-review postgres\n> Review PR #1234\n```\n\nIf you're on Claude Code and want the parallel review, also install `pr-review-toolkit`:\n\n```bash\nclaude plugin install pr-review-toolkit@claude-plugins-official\n```\n\n#### Load standards\n\nLoads all 22 connector standards into context. Run this before starting connector work so the agent actually knows what it's checking against.\n\n```\n> Load the DataHub standards\n> What are the connector standards?\n```\n\n## Installation\n\n### Quick install (any agent)\n\nThe [Skills CLI](https://github.com/vercel-labs/skills) detects your installed agents and sets things up:\n\n```bash\nnpx skills add datahub-project/datahub-skills\n```\n\nWorks with most agents including Claude Code, Cursor, Codex, Copilot, Gemini CLI, Windsurf, Cline, and Roo Code.\n\n### Platform-specific\n\n#### Claude Code\n\n```bash\n# Option A: Plugin install (gets you hooks, slash commands, multi-agent dispatch)\nclaude plugin install datahub-skills\n\n# Also install pr-review-toolkit for multi-agent reviews:\nclaude plugin install pr-review-toolkit@claude-plugins-official\n```\n\n```bash\n# Option B: Skills CLI (project-level, installs to .claude/skills/)\nnpx skills add datahub-project/datahub-skills -a claude-code\n```\n\nThen:\n\n```\n> Search for revenue tables in Snowflake\n> /datahub-search who owns the customer pipeline?\n> /datahub-enrich add description to orders table\n> /datahub-lineage what feeds into the revenue dashboard?\n> /datahub-quality find datasets with failing assertions\n> /datahub-setup verify my connection\n> /connector-review snowflake\n> /connector-planning duckdb\n```\n\n#### Cursor\n\n```bash\nnpx skills add datahub-project/datahub-skills -a cursor\n# Installs to .agents/skills/\n```\n\nCursor picks up skills from `.agents/skills/` automatically:\n\n```\n> Search DataHub for customer tables\n> Review my DataHub connector\n> Plan a connector for ClickHouse\n```\n\n#### GitHub Copilot\n\n```bash\nnpx skills add datahub-project/datahub-skills -a github-copilot\n# Installs to .agents/skills/\n```\n\nUse in Copilot Chat:\n\n```\n> Search the DataHub catalog for revenue data\n> Review my DataHub connector code\n> Help me plan a new connector for DuckDB\n```\n\n#### OpenAI Codex\n\n```bash\nnpx skills add datahub-project/datahub-skills -a codex\n# Installs to .agents/skills/\n```\n\n```\n> Find datasets owned by the data-eng team\n> Review the postgres connector against DataHub standards\n> Plan a connector for Snowflake\n```\n\n#### Gemini CLI\n\n```bash\nnpx skills add datahub-project/datahub-skills -a gemini-cli\n# Installs to .agents/skills/\n```\n\nVerify with `/skills list`, then:\n\n```\n> Who owns the revenue pipeline?\n> Review my DataHub connector\n> Plan a new connector for BigQuery\n```\n\n#### Windsurf\n\n```bash\nnpx skills add datahub-project/datahub-skills -a windsurf\n# Installs to .windsurf/skills/\n```\n\n```\n> Explore lineage for the orders table\n> Review my DataHub connector implementation\n> Plan a connector for Redshift\n```\n\n#### Manual install\n\n```bash\ngit clone https://github.com/datahub-project/datahub-skills.git\n\n# Catalog interaction skills\ncp -r datahub-skills/skills/datahub-search          your-project/.agents/skills/\ncp -r datahub-skills/skills/datahub-enrich           your-project/.agents/skills/\ncp -r datahub-skills/skills/datahub-lineage          your-project/.agents/skills/\ncp -r datahub-skills/skills/datahub-quality          your-project/.agents/skills/\ncp -r datahub-skills/skills/datahub-setup            your-project/.agents/skills/\ncp -r datahub-skills/skills/shared-references        your-project/.agents/skills/\ncp -r datahub-skills/skills/using-datahub            your-project/.agents/skills/\n\n# Connector development skills\ncp -r datahub-skills/skills/datahub-connector-planning   your-project/.agents/skills/\ncp -r datahub-skills/skills/datahub-connector-pr-review  your-project/.agents/skills/\ncp -r datahub-skills/skills/load-standards               your-project/.agents/skills/\n```\n\nEach skill directory is self-contained. The `standards` symlinks get dereferenced into real files on copy, so everything travels together. The catalog interaction skills reference `shared-references/` for CLI and MCP tool documentation.\n\n### What works where\n\n| Feature                     | Claude Code           | Cursor / Copilot / Codex / Gemini CLI / Windsurf |\n| --------------------------- | --------------------- | ------------------------------------------------ |\n| Catalog search              | Yes                   | Yes                                              |\n| Metadata enrichment         | Yes                   | Yes                                              |\n| Lineage exploration         | Yes                   | Yes                                              |\n| Data quality management     | Yes                   | Yes                                              |\n| Connection setup            | Yes                   | Yes                                              |\n| Planning workflow           | Yes                   | Yes                                              |\n| Load standards              | Yes                   | Yes                                              |\n| Review against standards    | Yes                   | Yes                                              |\n| Parallel multi-agent review | Yes (5 sub-agents)    | No (runs sequentially)                           |\n| Research agent delegation   | Yes (dedicated agent) | No (inline fallback)                             |\n| Slash commands              | Yes                   | No (use natural language instead)                |\n| SessionStart hooks          | Yes (via plugin)      | No                                               |\n\n## Commands (Claude Code only)\n\nOther platforms do the same things through natural language.\n\n### Catalog interaction\n\n| Command                     | What it does                                    |\n| --------------------------- | ----------------------------------------------- |\n| `/catalog-search [query]`   | Search the catalog and answer questions         |\n| `/catalog-enrich [entity]`  | Add or update metadata                          |\n| `/catalog-lineage [entity]` | Explore lineage and trace dependencies          |\n| `/catalog-quality [entity]` | Manage assertions, incidents, and subscriptions |\n| `/catalog-setup [task]`     | Set up connection and configure defaults        |\n\n### Connector development\n\n| Command                         | What it does                            |\n| ------------------------------- | --------------------------------------- |\n| `/connector-planning [source]`  | Plan a new connector                    |\n| `/connector-review [connector]` | Review connector code against standards |\n| `/load-standards`               | Load all 22 standards into context      |\n\n## Agents\n\n| Agent                        | What it does                                               |\n| ---------------------------- | ---------------------------------------------------------- |\n| `metadata-searcher`          | Fast sub-agent for executing catalog queries (Claude Code) |\n| `connector-researcher`       | Researches source systems before you write a connector     |\n| `connector-validator`        | Runs validation scripts and reports results                |\n| `comment-resolution-checker` | Checks whether PR review comments were actually addressed  |\n\n## Standards\n\n22 standards live in `standards/`, split into two groups:\n\n**Core (11):** main, api, sql, code_style, containers, lineage, patterns, performance, platform_registration, registration, testing\n\n**Source-type (11):** bi_tools, data_lakes, data_warehouses, identity_platforms, ml_platforms, nosql_databases, orchestration_tools, product_analytics, query_engines, sql_databases, streaming_platforms\n\n## Repo layout\n\n```\ndatahub-skills/\n├── .claude-plugin/\n│   ├── plugin.json\n│   └── marketplace.json\n├── skills/\n│   ├── datahub-search/              # Catalog search and discovery\n│   │   ├── SKILL.md\n│   │   ├── references/\n│   │   └── templates/\n│   ├── datahub-enrich/              # Metadata enrichment\n│   │   ├── SKILL.md\n│   │   ├── references/\n│   │   └── templates/\n│   ├── datahub-lineage/             # Lineage exploration\n│   │   ├── SKILL.md\n│   │   ├── references/\n│   │   └── templates/\n│   ├── datahub-quality/             # Data quality management\n│   │   ├── SKILL.md\n│   │   ├── references/\n│   │   └── templates/\n│   ├── datahub-setup/               # Connection setup and config\n│   │   ├── SKILL.md\n│   │   ├── references/\n│   │   └── templates/\n│   ├── datahub-connector-planning/  # Connector planning\n│   │   ├── SKILL.md\n│   │   ├── standards -> ../../standards\n│   │   ├── references/\n│   │   └── templates/\n│   ├── datahub-connector-pr-review/ # Connector review\n│   │   ├── SKILL.md\n│   │   ├── standards -> ../../standards\n│   │   ├── commands/\n│   │   ├── references/\n│   │   ├── scripts/\n│   │   └── templates/\n│   ├── load-standards/              # Load connector standards\n│   │   ├── SKILL.md\n│   │   └── standards -> ../../standards\n│   ├── shared-references/           # Shared CLI docs\n│   │   └── datahub-cli-reference.md\n│   └── using-datahub/              # Routing table (injected at session start)\n│       └── SKILL.md\n├── agents/\n│   ├── connector-researcher.md\n│   ├── comment-resolution-checker.md\n│   └── connector-validator.md\n├── commands/\n│   ├── catalog-search.md\n│   ├── catalog-enrich.md\n│   ├── catalog-lineage.md\n│   ├── catalog-quality.md\n│   ├── catalog-setup.md\n│   ├── connector-planning.md\n│   ├── connector-review.md\n│   └── load-standards.md\n└── standards/\n    ├── *.md (11 core)\n    └── source_types/*.md (11 source-type)\n```\n\nThe `standards` symlinks in each connector skill directory mean you can install a single skill and it brings its standards along. `npx skills add` dereferences these into real copies.\n\nThe catalog interaction skills share reference documents in `shared-references/` for CLI syntax, MCP tool signatures, and the DataHub entity model.\n\n## Contributing\n\nSee [CONTRIBUTING.md](CONTRIBUTING.md) for commit conventions and release process.\n\nWhere things live:\n\n- Catalog interaction skills: `skills/datahub-search/`, `skills/datahub-enrich/`, `skills/datahub-lineage/`, `skills/datahub-quality/`, `skills/datahub-setup/`\n- Shared references: `skills/shared-references/`\n- Connector standards: `standards/`\n- Review checklists: `skills/datahub-connector-pr-review/SKILL.md`\n- Planning steps: `skills/datahub-connector-planning/SKILL.md`\n- Agent prompts: `agents/`\n\n## License\n\nApache 2.0. See [LICENSE](LICENSE).\n",
  "bytes": 14398,
  "sha": "8f54202deefea625bd74d1878481c347616d60dbdf00d2174776feff7f85d8df",
  "repo_slug": "datahub-project/datahub-skills",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_datahub_project_datahub_skills_datahub_s_043ca96d/readme"
}