Verified Agent Skill record
yearning
Interact with the Yearning database change governance, automated and AI SQL review precheck, read-only SQL query, and SQL performance diagnostics platform through its /api/v4/* HTTP API, using the bundled zero-dependency scripts/yearning_api.py helper or standard HTTP requests. Covers querying pending approvals and orders, discovering granted flows and schemas, running SQL prechecks, managing change drafts, submitting, approving, executing, and rolling back change orders, running governed read-only SELECT queries, and diagnosing slow SQL statements.
Tasks
- List pending change orders awaiting your approval or execution using 'orders --pending-my-approval' or 'orders --pending-my-execution' via scripts/yearning_api.py, then inspect details with 'order <order_id>' and audit-reveal SQL with --reveal-sql when needed
- Run an automated and AI SQL review precheck against a change_review flow ('precheck --flow-id <flow_id> --sql ...'), then submit a governed change order bound to the passing run_id
- Execute governed read-only SELECT queries within an active query session, paging results with 'query-page' when next_cursor is non-null
- Run SQL performance diagnostics: collect a snapshot, view the 6-chart overview and top hotspot fingerprints, fetch statement detail, and request a desensitized AI root-cause report
- Approve, reject, comment on, withdraw/void, execute, or create a rollback draft for change orders
Inputs
- Environment variables or CLI flags: YEARNING_BASE_URL (default http://127.0.0.1:8080), YEARNING_USERNAME, YEARNING_PASSWORD or YEARNING_PASSWORD_FILE, YEARNING_AUTH_PROVIDER (local or ldap), optional YEARNING_SESSION_COOKIE and YEARNING_CSRF_TOKEN
- A concrete change_review flow_id and schema_name for precheck and order submission
- A datasource_id, schema_name, and active query session/grant for read-only queries
- SQL statements to precheck, submit as drafts/orders, or run as read-only SELECTs
- order_id, draft_id, report_id, statement_id, execution_id, and cursor values for pagination
Outputs
- Workbench profile and pending approval/draft counters ('me')
- Cursor-paginated lists of change orders and drafts, with metadata, timeline, findings, and comments per order
- Explicitly revealed (audited) order SQL and rollback SQL
- Precheck run results with status (ready/blocked/partial/failed/outdated) and a run_id for submission
- Read-only SELECT result pages with next_cursor for follow-up pages
- Diagnostics outputs: datasources status, 6-chart overview, 3-chart statement detail, parameterized SQL template, and AI optimization reports
- Rollback draft generated from an executed order
Limitations and checks
- Only SELECT statements are permitted in query sessions; all DDL/DML must go through precheck and submit-order
- Order and draft SQL is stored encrypted and returned as metadata only; SQL text requires an explicit audited sql-reveal call
- submit-order requires a valid run_id from precheck whose reviewed SQL matches the submitted SQL byte-for-byte
- The guide instructs stopping and asking the user to confirm when multiple candidate flows, schemas, or datasources match
- List endpoints use cursor pagination, so full traversal requires following the after cursor
- Confirm the target flow_id and schema_name are resolved via 'flows' and 'metadata' (or 'query-context' and 'query-metadata' for queries) before precheck, submission, or query execution—never guess
- Verify the reviewed SQL from the precheck run matches the submitted SQL byte-for-byte before submit-order
- Confirm user intent before decide-order, execute-order, or withdraw-or-void unless explicitly instructed to execute autonomously
Search
Find projects, verify facts, compare options, and filter by your requirements
Try a searchA click only fills the search box; you stay in control
Project Details
0