redshift-comment-mcp

Interactive walkthrough for an unfamiliar Redshift cluster — schema → table → column, picked by reading comments. Hands off to /redshift-profile. Use when user doesn't know where to start in a cluster. Do NOT use when user already knows the table.column (use /redshift-profile directly) or in non-interactive contexts. Triggers: /redshift-explore / browse Redshift / where do I look / 找 cluster / 從哪開始 / 探検 / ガイド付き探索.

Cross-table column search — finds every column whose name or comment matches a keyword across all tables in one (or all) schemas via one schema-wide MCP call per schema. Use when user wants to find FK / shared-key columns across many tables (e.g. before composing a JOIN), or to audit column-naming consistency. Do NOT use for a single known table (use search_columns directly), single-column lookup (use get_column_comment), or table-name search (use /redshift-grep-tables). Triggers: /redshift-grep-columns / find column / search columns across tables / where is foo column / 跨表找欄位 / 哪些表有 foo / カラム横断検索 / カラム名検索.

Cross-schema table search — finds every table whose name or comment matches a keyword across all schemas via one cluster-wide MCP call. Use when user is looking for a table by topic but unsure which schema (e.g. "where is the orders fact table"), or auditing table-naming consistency. Do NOT use when schema is already known (use search_tables directly), for column-level search (use /redshift-grep-columns), or for general schema browsing (use /redshift-explore). Triggers: /redshift-grep-tables / find table / search tables / which schema has / 哪個 schema 有 / 跨 schema 找表 / テーブル横断検索 / テーブル名検索.

Reconstruct ACTUAL table-to-table data flow from Redshift query history — catches ad-hoc, BI-tool (Tableau / Looker), and manual-fix usage dbt manifest cannot see. Read-only. Use when auditing who reads / writes a table, or before deprecating one. Do NOT use for dbt-internal lineage (use dbt manifest) or real-time monitoring. Triggers: /redshift-lineage-from-stl / actual lineage / who reads / Tableau usage / 實際 lineage / 誰在讀 / クエリ履歴.

Profile a Redshift column — cardinality, top-N values, null rate, min/max, plus comment. Read-only. Use when about to write CREATE TABLE or dbt schema.yml based on column assumptions, or to check whether a column is an enum. Do NOT use for full row counts (use execute_sql), schema/table search (use search_columns), or free-text columns where top-100 is noise. Triggers: /redshift-profile / profile column / distinct values / enum / 欄位分布 / カラムプロファイル / 値の分布.

Configure a Redshift connection profile for the redshift-comment-mcp plugin via chat — password stored in OS keychain via system dialog or terminal handoff (never in chat history), connection verified. Named form `/redshift-setup <name>` adds a non-default profile and offers to activate it. Use when setting up first Redshift cluster or adding another. Do NOT use for switching between already-configured profiles (use /redshift-switch-profile), password-only changes (use set-password CLI), or profile deletion (use delete-profile). Triggers: /redshift-setup / set up Redshift / add cluster / 設定 Redshift / 新增 cluster / 接続を設定 / プロファイル.

Switch the active Redshift profile by flipping the active-profile pointer file at ~/.config/redshift-comment-mcp/active-profile (no host / user / password re-entry). Verifies connection before declaring success. Single-profile users get a friendly bow-out pointing to /redshift-setup. Use when switching between 2+ already- configured Redshift clusters. Do NOT use for setting up a new profile (use /redshift-setup), changing passwords (use set-password CLI), or single-profile installs (skill bows out). Triggers: /redshift-switch-profile / switch cluster / change Redshift / 切換 cluster / 切換 profile / クラスタ切替 / 接続を切替.