Plugin: Data Professional (Data Team Catalog)

Cursor plugin for the data team: Data Analyst, Data Scientist, and Data Engineer agents, skills, and commands. Supports optional subagent usage so heavy steps (discovery, lineage, EXPLAIN) run in isolated context without polluting the main task.

Contents

• Agents: agents/data-analyst.md, data-scientist.md, data-engineer.md — when to use which role.
• Skills: skills/data-analyst/, data-scientist/, data-engineer/ — each has SKILL.md with frontmatter. Data Analyst wraps the full pro-data-analyst workflow and optional subagent usage.
• Commands: commands/da-data-brief.md, da-discover-tables.md, da-run-lineage.md, da-explain-query.md, da-full-analysis.md, da-skip-checkpoints.md (analyst); commands/dk-map-domain.md, dk-sync-domain-diagram.md (knowledge map) — some support a "subagent" option.
• Docs: docs/AGENTS.md, SKILLS.md, COMMANDS.md, SUBAGENT-USAGE.md — catalog and I/O contract for subagents.
• Prompts: prompts/discovery.md, lineage.md, explain-and-run.md — templates to pass into subagent tasks (fill placeholders before use).

Installation

Install as a Cursor plugin (e.g. from repo or local path). Ensure .cursor-plugin/plugin.json is present; Cursor will discover rules/, skills/, agents/, commands/, and hooks/ by default.

Usage

1. Data Analyst: Invoke by intent ("revenue report", "find tables for X", "lineage for table Y") or use a command (e.g. da-full-analysis, da-discover-tables). Say "run discovery in subagent" or "lineage via subagent" to delegate those steps.
2. Data Scientist: Invoke by intent ("correlation analysis", "train model", "ML pipeline"). Uses the data-scientist skill wrapper (pandas, ML, notebooks).
3. Data Engineer: Invoke by intent ("ETL from X", "DWH table design", "data quality check"). Uses the data-engineer skill wrapper (ETL, DWH, orchestration).

Subagent usage

When the user asks for discovery, lineage, or EXPLAIN to run in a subagent:

• The main agent calls mcp_task with the appropriate prompt from prompts/. The prompt must be self-contained (include script paths and db alias); the subagent runs in isolated context.
• The subagent returns a summary (tables/columns, lineage, or EXPLAIN + sample). The main agent reads it and continues with the next phase or checkpoint.
• All checkpoints (1–4) remain with the main agent.

See docs/SUBAGENT-USAGE.md for the full contract and docs/COMMANDS.md for which commands support the subagent option.

Dependencies

• Pro-data-analyst (full skill): For the full 7-phase analyst workflow, the skill is expected at ~/.claude/skills/pro-data-analyst/SKILL.md (or equivalent). The plugin’s skills/data-analyst/SKILL.md is a wrapper that references it and adds subagent usage.
• Scripts and data: If your project uses scripts/ (e.g. search_documents.py, search_schema.py, explain_query.py, run_query_safe.py, search_procedures.py), documents/, and scenarios/, ensure paths in the prompts point to the correct workspace. Subagent prompts use placeholders (e.g. {{WORKSPACE_ROOT}}) that the main agent must fill.

License

See repository or plugin manifest.