Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Migration Parity & Exceptions Verifier

    1

    Compare CSV/JSON by key. Find missing rows, duplicates and value differences with local reports.

    $14.99

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Migration Parity & Exceptions Verifier

    Migration Parity & Exceptions Verifier

    Example session with this skill installed

    Compare the bundled synthetic equal-totals exports using config.json. Run locally in a fresh directory, preserve inputs, and report missing keys, duplicates and value differences. Do equal totals prove parity?

    • Read your context and instructions
    • Compiled the migration parity
    • Generated the data export

    Actual CLI evidence — synthetic fixtures

    The tool ran locally. This explanation is manually written from its generated artifacts, not a recorded AI session.

    python3 migration-parity-verifier/examples/run_demo.py --out /tmp/parity-demo-new
    

    The runner exited 0 after verifying all four expected comparisons:

    FixtureComparison exitExceptions
    parity00
    equal-totals13
    duplicates11
    mapped JSON-to-CSV00

    Equal-totals result

    EXCEPTIONS, exit 1. Left: 2 data rows. Right: 2 data rows.

    IssueCount
    duplicate_left0
    duplicate_right0
    missing_left1
    missing_right1
    value_difference1

    Left has A100=40.00, B200=60.00. Right has A100=45.00, C300=55.00. Both total 100.00. A100 amount differs; B200 is missing on the right; C300 is missing on the left. Equal totals do not establish parity.

    The complete exception CSV, Markdown report and hash manifest ship in migration-parity-verifier/examples/demo-output/equal-totals/. New runs write the same three report types in your fresh output directory.

    Key: id (text). Compared fields: id and amount (exact decimal). Null tokens: none. Trimming: off. Naive times: reject. Tool version: 1.0.0.

    Review locally

    Inspect exceptions.csv using its documented JSON-encoded cell format. manifest.json records original input/config/script hashes, normalized rules and limits. Repeated reports are byte-identical with identical inputs, config, tool, limits and environment. Inputs were unchanged.

    This configured result does not approve the whole migration or cutover. No data was repaired or transmitted.

    exceptions.csv

    CSV · data export

    Generated

    Example file from a real run - the skill writes it into your workspace.

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

    About this skill

    Find the records behind the totals

    Compare local CSV or JSON exports by explicit business keys. Find missing rows, duplicate keys and changed values, then keep reproducible exception reports for review.

    USD 14.99 — one-time purchase.

    What you get

    • A reusable SKILL.md workflow for agreeing the comparison contract and reviewing evidence.
    • A deterministic Python 3.10+ tool using only the standard library.
    • Composite keys, left/right field mappings, exact decimals, configurable null tokens and documented timezone behavior.
    • exceptions.csv, a Markdown report and a manifest with input, configuration and script SHA-256 hashes.
    • Synthetic examples, installation instructions, a worked demo and automated tests.

    Equal totals can hide different records

    The synthetic source has A100 = 40.00 and B200 = 60.00. The target has A100 = 45.00 and C300 = 55.00. Both totals are 100.00, yet the tool finds three exceptions: one changed amount, one missing-right key and one missing-left key.

    Explicit comparison rules

    Choose fields as text, decimal, timestamp or boolean. Decimal values compare exactly: 1.00 equals 1, while 0.10 differs from 0.1001. No tolerance, automatic rounding, currency conversion or unit scaling. Null tokens are opt-in; JSON null differs from empty text by default. Offset-bearing timestamps compare as UTC instants. Naive timestamps require an explicit timezone policy; ambiguous or nonexistent IANA local times are rejected.

    Duplicate keys are reported. The tool does not guess which rows to pair. Unconfigured scalar fields are outside scope; nested JSON values are rejected.

    Run locally and review

    Agree keys, mappings and normalization rules, then run the script against authorized local files in a fresh output directory. Inspect the exception CSV, Markdown summary and manifest locally. Repeat using identical inputs, configuration, tool, limits and runtime/timezone environment.

    The tool has no database connector, network calls, telemetry or automatic repair. It never changes input data. The AI workflow defaults to metadata and counts. Reports may contain configured source values and should stay local unless you explicitly authorize sharing them. Your AI host privacy behavior is separate.

    Requirements and limits

    Python 3.10+ and a local shell. Guided skill use requires a host that reads local SKILL.md and supports tool execution; AI usage is separate. CSV and flat JSON row arrays only. Processing is in memory, with defaults of 25 MiB per file and 100,000 rows per input, configurable through CLI limits. No large-scale performance benchmark is claimed.

    Results cover the supplied exports, keys, fields and rules. They do not establish export completeness, referential integrity, application behavior or overall cutover readiness. A zero-exception result is evidence for the configured comparison; human owners decide next steps. No consulting or managed migration service is included.

    CSV and flat JSON rows only. No Excel, nested JSON, joins, fuzzy matching, database access or repair. In-memory: defaults 25 MiB/file and 100,000 rows/input. Larger limits need validation; no scale benchmark. IANA rules depend on host timezone data. Reports may contain source values. A clean result covers supplied exports and configured rules only, not completeness, application behavior or cutover readiness. AI privacy/costs are separate.

    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

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    Requires

    Python 3.10+
    Python

    Permissions

    Terminal / Shell
    Read Files
    Write Files

    Reads only authorized local CSV/JSON exports and the reviewed comparison configuration. Writes exception CSV, Markdown, and manifest files to a new local output directory. No database connections, automatic repair, network calls, or telemetry. Reports may contain configured values; keep them local. AI-agent usage and privacy depend on the host you choose.

    Listed1 month ago

    What's inside

    Frequently Asked Questions

    No. It compares authorized local exports. It has no database connector, network upload or repair feature and never changes input data. A person decides how to resolve exceptions.

    Yes. CSV or flat JSON row arrays can be used on either side. Configure explicit keys and left/right field mappings. Nested JSON values are rejected. Unconfigured scalar fields are outside scope.

    Decimals are exact with no tolerance or rounding. Offset timestamps compare as UTC instants; naive times need explicit rules. Ambiguous/nonexistent IANA times reject. Null tokens are opt-in.

    Duplicate keys produce exceptions. The tool never picks one row arbitrarily, so ambiguous keys cannot silently pass. Investigate duplicates and choose keys that reflect record identity.

    Python 3.10+; no extra packages. In-memory defaults: 25 MiB/file and 100,000 rows/input. Limits are configurable; validate your workload. No streaming or scale benchmark.

    New in Data & Databases