MCP-Airflow-API

Returns the MCP prompt template (full, headings, or specific section). Args: section: Section number or keyword (optional) mode: 'full', 'headings', or None (optional)

[Tool Role]: Lists all DAGs registered in the Airflow cluster with pagination support.

Args: limit: Maximum number of DAGs to return (default: 20) - For small queries: use default 100 - For large environments: use 500-1000 to get more DAGs at once - Maximum recommended: 1000 (to avoid API timeouts) offset: Number of DAGs to skip for pagination (default: 0) - Use 0 for first page - Use limit*page_number for subsequent pages - Example: offset=100 for page 2 when limit=100

Pagination Examples: - First 20 DAGs: list_dags() - Next 20 DAGs: list_dags(limit=20, offset=20)
- Page 3 of 50 DAGs each: list_dags(limit=50, offset=100) - All DAGs at once: list_dags(limit=1000)

Use total_entries from response to determine if more pages exist: - has_more_pages = (offset + limit) < total_entries - next_offset = offset + limit - remaining_count = total_entries - (offset + limit)

Returns: Dict containing: - dags: List of DAG objects with comprehensive info (dag_id, dag_display_name, is_active, is_paused, description, schedule_interval, max_active_runs, max_active_tasks, owners, tags, next_dagrun info, last_parsed_time, has_import_errors, timetable_description) - total_entries: Total number of DAGs in Airflow (for pagination calculation) - limit: Requested limit (echoed back) - offset: Requested offset (echoed back) - returned_count: Actual number of DAGs returned in this response - has_more_pages: Boolean indicating if more pages are available - next_offset: Suggested offset for next page (if has_more_pages is True)

[Tool Role]: Retrieves detailed information for a specific DAG.

Args: dag_id: The DAG ID to get details for

Returns: Comprehensive DAG details: dag_id, schedule_interval, start_date, owners, tags, description, etc.

[Tool Role]: Retrieves detailed information for multiple DAGs in batch with get_dag() level detail plus latest run information.

This tool combines list_dags() filtering with get_dag() detailed information retrieval, providing comprehensive DAG details AND latest execution information for multiple DAGs in a single response. Each DAG entry includes both static configuration details and dynamic runtime information.

Args: limit: Maximum number of DAGs to process (default: 100) - Use higher values (500-1000) for large environments - Ignored when fetch_all=True offset: Number of DAGs to skip for pagination (default: 0) fetch_all: If True, fetches all DAGs regardless of limit/offset (default: False) id_contains: Filter DAGs by ID containing this string (optional) name_contains: Filter DAGs by display name containing this string (optional) is_active: Filter by active status - True/False (optional) is_paused: Filter by paused status - True/False (optional)

Usage Examples: - All unpaused DAGs with full details and latest runs: get_dags_detailed_batch(fetch_all=True, is_paused=False) - Active, unpaused DAGs only: get_dags_detailed_batch(is_active=True, is_paused=False) - DAGs containing "example": get_dags_detailed_batch(id_contains="example", limit=50) - Paginated batch: get_dags_detailed_batch(limit=100, offset=200)

Returns: Dictionary containing: - dags_detailed: List of detailed DAG objects with: * All get_dag() fields (dag_id, schedule_interval, start_date, owners, tags, etc.) * latest_dag_run: Most recent execution information (run_id, state, start_date, end_date, etc.) - total_processed: Number of DAGs successfully processed - total_available: Total number of DAGs matching initial filters - processing_stats: Success/failure counts and error details - applied_filters: Summary of filters applied - pagination_info: Current page info and remaining counts

[Tool Role]: Lists all currently running DAG runs in the Airflow cluster.

Returns: List of running DAG runs with comprehensive info: dag_id, run_id, state, execution_date, start_date, end_date, data_interval_start, data_interval_end, run_type, conf, external_trigger, dag_display_name

[Tool Role]: Lists all recently failed DAG runs in the Airflow cluster.

Returns: List of failed DAG runs with comprehensive info: dag_id, run_id, state, execution_date, start_date, end_date, data_interval_start, data_interval_end, run_type, conf, external_trigger, dag_display_name

[Tool Role]: Triggers a new DAG run for a specified Airflow DAG.

Args: dag_id: The DAG ID to trigger

Returns: Minimal info about triggered DAG run: dag_id, run_id, state, execution_date, start_date, end_date

[Tool Role]: Pauses the specified Airflow DAG (prevents scheduling new runs).

Args: dag_id: The DAG ID to pause

Returns: Minimal info about the paused DAG: dag_id, is_paused

[Tool Role]: Unpauses the specified Airflow DAG (allows scheduling new runs).

Args: dag_id: The DAG ID to unpause

Returns: Minimal info about the unpaused DAG: dag_id, is_paused

[Tool Role]: Retrieves the task dependency graph structure for a specific DAG.

Args: dag_id: The DAG ID to get graph structure for

Returns: DAG graph with tasks and dependencies: dag_id, tasks, dependencies

[Tool Role]: Lists all tasks for a specific DAG.

Args: dag_id: The DAG ID to get tasks for

Returns: List of tasks with detailed task information: dag_id, tasks, total_tasks

[Tool Role]: Retrieves the source code for a specific DAG.

Args: dag_id: The DAG ID to get source code for

Returns: DAG source code: dag_id, file_token, source_code

[Tool Role]: Lists event log entries with optional filtering.

Args: dag_id: Filter by DAG ID (optional) task_id: Filter by task ID (optional) run_id: Filter by run ID (optional) limit: Maximum number of log entries to return (default: 20, increased from 20 for better coverage) offset: Number of entries to skip (default: 0)

Returns: List of event logs: event_logs, total_entries, limit, offset, pagination metadata

[Tool Role]: Retrieves a specific event log entry by ID.

Args: event_log_id: The event log ID to retrieve

Returns: Single event log entry: event_log_id, when, event, dag_id, task_id, run_id, etc.

[Tool Role]: Retrieves event count summary for all DAGs.

Returns: Summary of event counts by DAG: dag_summaries, total_dags, total_events

[Tool Role]: Lists import errors with optional filtering.

Args: limit: Maximum number of import errors to return (default: 20, increased from 20 for better coverage) offset: Number of entries to skip (default: 0)

Returns: List of import errors: import_errors, total_entries, limit, offset, pagination metadata

[Tool Role]: Retrieves a specific import error by ID.

Args: import_error_id: The import error ID to retrieve

Returns: Single import error: import_error_id, filename, stacktrace, timestamp

[Tool Role]: Retrieves import error summary for all DAGs.

Returns: Summary of import errors by filename: import_summaries, total_errors, affected_files

[Tool Role]: Retrieves run duration statistics for a specific DAG.

Args: dag_id: The DAG ID to get run durations for limit: Maximum number of recent runs to analyze (default: 50, increased from 10 for better statistics)

Returns: DAG run duration data: dag_id, runs, statistics

[Tool Role]: Retrieves task duration information for a specific DAG run.

Args: dag_id: The DAG ID to get task durations for run_id: Specific run ID (if not provided, uses latest run)

Returns: Task duration data: dag_id, run_id, tasks, statistics

[Tool Role]: Retrieves calendar/schedule information for a specific DAG.

Args: dag_id: The DAG ID to get calendar info for start_date: Start date for calendar range (YYYY-MM-DD format, optional) end_date: End date for calendar range (YYYY-MM-DD format, optional) limit: Maximum number of DAG runs to return (default: 20, was hardcoded at 50)

Returns: DAG calendar data: dag_id, schedule_interval, runs, next_runs

[Tool Role]: Get the health status of the Airflow webserver instance.

Returns: Health status information including metadatabase and scheduler status

[Tool Role]: Get version information of the Airflow instance.

Returns: Version information including Airflow version, Git version, and build info

[Tool Role]: List all pools in the Airflow instance.

Args: limit: Maximum number of pools to return (default: 20) offset: Number of pools to skip for pagination (default: 0)

Returns: List of pools with their configuration and usage information

[Tool Role]: Get detailed information about a specific pool.

Args: pool_name: The name of the pool to retrieve

Returns: Detailed pool information including slots usage and description

[Tool Role]: Lists task instances across all DAGs or filtered by specific DAG with comprehensive filtering options.

IMPORTANT: When users provide natural language dates, calculate relative dates using the current server time context (internally via get_current_time_context):

• "yesterday" = current_date - 1 day
• "last week" = current_date - 7 days to current_date - 1 day
• "last 3 days" = current_date - 3 days to current_date
• "today" = current_date

Args: dag_id: Filter by DAG ID (optional) dag_run_id: Filter by DAG run ID (optional) execution_date_gte: Filter by execution date greater than or equal to (ISO 8601 format with timezone, e.g., '2024-01-01T00:00:00Z', optional) execution_date_lte: Filter by execution date less than or equal to (ISO 8601 format with timezone, e.g., '2024-01-01T23:59:59Z', optional) start_date_gte: Filter by start date greater than or equal to (ISO 8601 format with timezone, optional) start_date_lte: Filter by start date less than or equal to (ISO 8601 format with timezone, optional) end_date_gte: Filter by end date greater than or equal to (ISO 8601 format with timezone, optional) end_date_lte: Filter by end date less than or equal to (ISO 8601 format with timezone, optional) duration_gte: Filter by duration greater than or equal to (seconds, optional) duration_lte: Filter by duration less than or equal to (seconds, optional) state: Filter by task state (queued, running, success, failed, up_for_retry, up_for_reschedule, upstream_failed, skipped, deferred, scheduled, removed, restarting, optional) pool: Filter by pool name (optional) queue: Filter by queue name (optional) limit: Maximum number of task instances to return (default: 20) offset: Number of task instances to skip for pagination (default: 0)

Returns: List of task instances with comprehensive information: task_instances, total_entries, limit, offset

[Tool Role]: Retrieves detailed information about a specific task instance.

Args: dag_id: The DAG ID containing the task dag_run_id: The DAG run ID containing the task instance task_id: The task ID to retrieve details for

Returns: Detailed task instance information including execution details, state, timing, and configuration

[Tool Role]: Lists task instances in batch with multiple filtering criteria for bulk operations.

Relative date filters (if provided) are resolved against the server's current time.

Args: dag_ids: List of DAG IDs to filter by (optional) dag_run_ids: List of DAG run IDs to filter by (optional) task_ids: List of task IDs to filter by (optional) execution_date_gte: Filter by execution date greater than or equal to (ISO format, optional) execution_date_lte: Filter by execution date less than or equal to (ISO format, optional) start_date_gte: Filter by start date greater than or equal to (ISO format, optional) start_date_lte: Filter by start date less than or equal to (ISO format, optional) end_date_gte: Filter by end date greater than or equal to (ISO format, optional) end_date_lte: Filter by end date less than or equal to (ISO format, optional) duration_gte: Filter by duration greater than or equal to (seconds, optional) duration_lte: Filter by duration less than or equal to (seconds, optional) state: List of task states to filter by (optional) pool: List of pool names to filter by (optional) queue: List of queue names to filter by (optional)

Returns: Batch list of task instances with filtering results: task_instances, total_entries, applied_filters

[Tool Role]: Lists extra links for a specific task instance (e.g., monitoring dashboards, logs, external resources).

Args: dag_id: The DAG ID containing the task dag_run_id: The DAG run ID containing the task instance task_id: The task ID to get extra links for

Returns: List of extra links with their URLs and descriptions: task_id, dag_id, dag_run_id, extra_links

[Tool Role]: Retrieves logs for a specific task instance and its try number with content and metadata.

Args: dag_id: The DAG ID containing the task dag_run_id: The DAG run ID containing the task instance task_id: The task ID to get logs for try_number: The try number for the task instance (default: 1) full_content: Whether to return full log content or just metadata (default: False) token: Pagination token for large logs (optional)

Returns: Task instance logs with content and metadata: task_id, dag_id, dag_run_id, try_number, content, metadata

[Tool Role]: Lists all variables stored in Airflow.

Args: limit: Maximum number of variables to return (default: 20) offset: Number of variables to skip for pagination (default: 0) order_by: Order variables by field (key, description) (default: key)

Returns: List of variables with their keys, values, and descriptions

[Tool Role]: Retrieves a specific variable by its key from Airflow.

Args: variable_key: The key of the variable to retrieve

Returns: Variable information including key, value, and description

[Tool Role]: Lists XCom entries for a specific task instance.

Args: dag_id: The DAG ID dag_run_id: The DAG run ID
task_id: The task ID limit: Maximum number of entries to return (default: 20) offset: Number of entries to skip (default: 0)

Returns: Dictionary containing XCom entries with key, value, timestamp, and other metadata

[Tool Role]: Retrieves a specific XCom entry for a task instance.

Args: dag_id: The DAG ID dag_run_id: The DAG run ID task_id: The task ID
xcom_key: The XCom key to retrieve map_index: Map index for mapped tasks (default: -1 for non-mapped)

Returns: Dictionary containing the specific XCom entry with full value and metadata