메뉴
Build and run dbt transformations in Dalgo — the transform canvas, operations, model materialization, dbt runs, and publishing to git. Use when a user asks to run dbt, build or edit models, or modify the transform canvas.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Work with Dalgo data ingestion — Airbyte sources, connections, stream catalogs, and sync history. Use when a user asks to sync a source, check whether a sync ran, or inspect what a connection syncs.
Route Dalgo MCP tool workflows. Use when a request involves Dalgo data, pipelines, syncs, transforms, dashboards, charts, or reports, to pick the right domain skill and stay on MCP tools.
Run, monitor, create, and manage Dalgo orchestration pipelines (Prefect deployments), including the trigger-poll-verify loop. Use whenever a user asks to run a pipeline or check whether one succeeded.
Diagnose Dalgo failures — failed pipeline runs, sync errors, dbt errors — and handle notifications. Use when a user asks why something failed, what went wrong, or what their notifications say.
Create and manage Dalgo charts, dashboards, and reports. Use when a user asks to visualize data, build or modify a dashboard, or snapshot data into a report.
Explore the Dalgo data warehouse — schemas, tables, columns, row counts, and sample data. Use for questions about what data exists or what values a table contains.
| name | dalgo-transforms |
| description | Build and run dbt transformations in Dalgo — the transform canvas, operations, model materialization, dbt runs, and publishing to git. Use when a user asks to run dbt, build or edit models, or modify the transform canvas. |
Use for dbt and transform-canvas workflows through Dalgo MCP tools.
dalgo_acquire_canvas_lock before any modification, dalgo_release_canvas_lock after — release even if a step fails, or other users stay locked out.dalgo_run_dbt is fire-and-forget (queued via Celery): a 200 response means queued, not succeeded. Wait 30–60 seconds (longer for big projects) before checking outcomes; never report success from the trigger response.dalgo_sync_sources updates the dbt project's source definitions from the warehouse — it does NOT sync raw data from external sources. For that, route to dalgo-ingestion.dalgo_publish_changes (commits and pushes to git) and check dalgo_get_git_status first so you can say what will be committed.dalgo_run_dbt with empty run_params. Specific model: {"select": "model_name"}; model plus downstream: {"select": "model_name+"}.dalgo_get_transform_graph and describe output models in plain English, not node IDs.dalgo_sync_sources first, then re-run; if raw data itself is missing, route to dalgo-ingestion.dalgo_add_source_to_canvas → dalgo_create_operation chain → dalgo_terminate_chain to materialize → unlock. Use dalgo_get_data_types for cast operations.dalgo_get_transform_graph, dalgo_get_node_details / dalgo_get_node_columns for specifics.dalgo_run_dbt → wait → verify via dalgo_get_sources_models and dalgo_get_git_status.