T-SQL to Databricks Migration Auditor

    1

    Scan SQL Server T-SQL for what breaks on Databricks, get risk-ranked fixes, MERGE rewrites and reconciliation queries.

    Free

    1 useSecurity scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    T-SQL to Databricks Migration Auditor

    Example session with this skill installed

    Audit sql/usp_LoadDailyOrders.sql for our Databricks migration and propose the Databricks version.

    • Read your context and instructions
    • Compiled the t-sql to databricks

    (This is the scanner's real output on the bundled example examples/sample_proc.sql, plus the skill's rewrite. The full table is shortened here.)

    
    # T-SQL -> Databricks scan
    Files scanned: 1 | findings: 24 (high 9, medium 6, low 9)
    
    | high   | :1  | Stored procedure       | Unity Catalog SQL procedure or notebook/job task |
    | high   | :5  | TRY/CATCH              | SQL scripting handler or PySpark try/except |
    | high   | :6  | Explicit transaction   | single MERGE or stage + swap |
    | high   | :9  | CONVERT with style 120 | date_format(d,'yyyy-MM-dd') |
    | high   | :11 | #temp table            | CREATE OR REPLACE TEMP VIEW / CTE |
    | medium | :16 | UPDATE ... FROM join   | MERGE INTO |
    | low    | :12 | NOLOCK hint            | remove |
    
    

    Effort: 1 object, bucket = redesign (procedure + transaction + temp table).

    Proposed Databricks SQL

    CREATE OR REPLACE TEMP VIEW recent AS
    SELECT o.OrderID,
           COALESCE(c.`Customer Name`, 'Unknown') AS CustomerName,
           date_format(o.OrderDate, 'yyyy-MM-dd') AS OrderDay,
           concat(c.FirstName, ' ', c.LastName) AS FullName,   -- VERIFY: NULL names give NULL, same as T-SQL '+'
           o.Amount
    FROM main.sales.orders o
    JOIN main.sales.customers c ON c.CustomerID = o.CustomerID  -- VERIFY: case-sensitive compare if keys are strings
    WHERE o.OrderDate >= dateadd(DAY, -1, :run_date);
    -- VERIFY: the original TOP 1000 had no ORDER BY, so it was non-deterministic. It's dropped here; confirm the intent.
    
    MERGE INTO main.sales.fact_orders t
    USING recent r ON r.OrderID = t.OrderID
    WHEN MATCHED THEN UPDATE SET t.Amount = r.Amount;
    -- A single MERGE is atomic on Delta, so the BEGIN TRAN/COMMIT wrapper isn't needed.
    

    Then come row-count, sum(Amount) and hash-diff queries for fact_orders on both sides. (main.sales is a placeholder; the skill asks for your catalog mapping.)

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Migrating SQL Server to Databricks? Most of the pain isn't syntax. It's the code that still runs after conversion but returns different numbers: DATEDIFF boundary counting, case-sensitive joins, CONVERT style codes, time zones, and multi-statement transactions that Delta won't treat as one unit.

    This skill gives your agent a repeatable migration-audit workflow:

    • Scans .sql files and folders with a bundled, read-only Python scanner (stdlib only, no installs). It covers 41 T-SQL patterns and reports file:line evidence and severity.
    • Triages each object as mechanical, review, or redesign, so you can scope the work honestly.
    • Rewrites high-risk objects set-based: cursors and WHILE loops become window functions, UPDATE…FROM becomes MERGE INTO, #temp tables become temp views, and dbo maps to Unity Catalog. Every behaviour change is marked with -- VERIFY:.
    • Reconciles with matching source and target queries for row counts, key uniqueness, null and sum profiles, and SHA-256 row hashes.

    Built by a data engineer who has done SQL Server → Databricks Delta migrations in production.

    What it won't do: connect to any database, run SQL, edit your files, or convert SSIS packages.

    Does not connect to any database, run SQL, edit your files, or convert SSIS packages (SSIS/SSRS out of scope; it only lists embedded SQL). The scanner is regex-based and can give false positives (e.g. + in arithmetic flagged as concatenation); the agent confirms each finding in context.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    1 use

    Times used to date

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    • Free with an Agensi account

    Requires

    Python 3

    Permissions

    Terminal / Shell
    Read Files

    Read: your .sql files Bash: `python scripts/scan_tsql.py` (Python 3 standard library, read-only, no network, no file writes) No credentials, database connections or write access

    Listed1 month ago

    What's inside

    Frequently Asked Questions

    No. It's read-only, and it generates queries for you to run.

    No. Use those for bulk conversion. This skill audits the code before and after conversion, explains the risky parts, and gives you reconciliation checks.

    No. It's regex-based and can give false positives (for example, `+` in arithmetic flagged as concatenation). The agent confirms each finding in context.

    Out of scope. It only lists any SQL embedded in them.

    Any agent that supports the SKILL.md standard and can run Python 3 (Claude Code, Cursor, Codex, Gemini CLI).

    New in Data & Databases