Webflow CMS Locale Inventory Check

What does Webflow CMS Locale Inventory Check do?

Compare expected item IDs and CMS locale IDs with supplied JSON or CSV inventory snapshots. This is an offline data-processing Actor for Webflow CMS delivery checks. It does not connect to Webflow, fetch pages, translate text, publish content, or decide which locales your business needs.

Why use an inventory receipt?

A connector can finish its export while a requested locale variant is missing. Comparing totals alone can also hide a different item occupying the missing item's place. This check compares exact collection/item/locale keys and keeps intentional locale exclusions explicit. It can be repeated after an import, localization handoff or agency batch delivery. Apify can expose the same input/output through its API and integrations; this Actor needs no proxy or external AI service.

The receipt is evidence about supplied inventory, not an assurance that a live website or translation is correct. Existing TMS/native workflows may already meet your needs.

Input contract

Use Demo for the built-in five-row synthetic exercise. Empty API input also selects this exercise. Demo fields may equal the supplied fixture; changing them requires selecting Check. Check mode requires both expected and captures. One missing document never silently receives unrelated sample data. See the Input tab and fixtures/actor-input.json.

expected has schemaVersion: 1 and an expected array. Each row contains collectionId, itemId, cmsLocaleId, and boolean required. All variants of the same item share the base item ID; CMS locale IDs are distinct from page locale IDs. required: false means intentionally not required, not that this variant must be deleted.

Minimal synthetic check (one present variant):

{
  "mode": "check",
  "expected": {
    "schemaVersion": 1,
    "expected": [{"collectionId": "sample-collection", "itemId": "sample-item", "cmsLocaleId": "sample-fr", "required": true}]
  },
  "captures": {
    "schemaVersion": 1,
    "snapshots": [{
      "collectionId": "sample-collection", "cmsLocaleId": "sample-fr",
      "format": "json", "complete": true, "unfiltered": true,
      "pages": [{"items": [{"id": "sample-item", "cmsLocaleId": "sample-fr"}],
                 "pagination": {"offset": 0, "limit": 100, "total": 1}}]
    }]
  }
}

captures has schemaVersion: 1 and a snapshots array. Each snapshot covers one collection and one CMS locale. complete: true and unfiltered: true are required declarations; their safe default is unknown. Those declarations are not independently authenticated. Do not mark a filtered or partially exported inventory complete.

JSON snapshots contain format: "json" and pages holding the Webflow response shape: items plus pagination.limit, offset, and total. All pages must be contiguous, totals consistent, and variants unique and tagged with the declared CMS locale. Missing pages, duplicate scopes, mismatched locales and malformed captures make the entire affected scope unknown. A missing scope is unknown, even when the expected item exists in another locale.

CSV snapshots contain format: "csv", csvText, complete: true, unfiltered: true, integer total, and an explicit columns map. Map existing headers to id and cmsLocaleId, with optional isDraft, isArchived, and lastPublished. No header or locale inference occurs. Per-locale native exports might require a caller-supplied locale column after verifying their source. This tool cannot certify that preparation. See fixtures/captures-csv.json.

Limits: entire Actor input 2 MB, each CLI input file 2 MB, 10,000 expectations, 100 snapshot scopes, 10,000 observed items per scope, and JSON pages of at most 100 items. No credential inputs exist. Remove fieldData, private content and credentials before preparing an authorized snapshot; IDs themselves may also be confidential.

Output

The dataset has one row per expectation. Download it in JSON, HTML, CSV or Excel through Apify.

Field	Type	Meaning
collectionId, itemId, cmsLocaleId	string	Exact declared identity
required	boolean	Caller explicitly requests this variant
presence	string	present, absent or unknown
result	string	Required presence result, or not_required
reason	string	Coverage explanation
publicationObservation	object/null	Draft/archive flags and historical lastPublished
inputKind	string	synthetic_demo or supplied_snapshot
receiptStatus	string	Comparison status; in PPE mode inspect the final receipt

Key-value outputs: RECEIPT (JSON report), REPORT.md (deterministic human report), STATE (completion and delivery checkpoint), and ERROR on failure. Aborted or failed delivery must not be treated as complete; inspect STATE.completed. writtenRows counts acknowledged pushes. On interruption it is a lower bound: a row may reach the dataset before its acknowledgement returns. pendingRows can identify one in-flight push, whose delivery is uncertain. Existing rows can remain after an interrupted run; reports and a checkpoint do not replace inspection of that partial dataset.

Presence and publication are separate. An item with isDraft: true is still present in a staged capture. A historical lastPublished does not prove what is live now. Missing flags stay null. Translation completeness, language accuracy, links, images, publishing freshness and live HTTP behavior are excluded.

Local use and verification

The pure core needs only Python 3.13 and the standard library:

python -B -m unittest -v test_receipt.py
python receipt.py fixtures/expected.json fixtures/captures.json NEW_OUTPUT_DIRECTORY

The CLI creates receipt.json, receipt.md, and source-hashes.json without changing input files or overwriting an existing output directory. Exit 0 means no required variant is missing in supplied complete snapshots; exit 2 means absent/unknown requires review; exit 3 means invalid input and no completed receipt. The provided exercise intentionally returns exit 2: two present, one absent, one unknown, and one not required.

Run the Actor locally with apify run --user-agent apify-agent-skills/apify-actor-development. The default exercise emits five synthetic rows without visiting Webflow or any destination website. Actor execution uses Apify storage for its dataset and reports.

Cost and distribution

The pay-per-event contract is $0.002 per saved inventory row ($2 per 1,000), using only the custom locale-variant event. There is no second dataset-item or startup fee. A row with an absent, unknown or intentionally not-required result is still a saved result and can incur this fee. The five-row synthetic demonstration can therefore cost $0.01 in a paid run; it is not a free customer audit. Invalid top-level input produces no result rows. The actor validates the exact sole positive event price before writing paid results.

Your maximum event charge limits delivered rows. A $0.006 event budget delivers at most three rows; the receipt becomes BUDGET_LIMITED, contains only acknowledged charged rows, and does not report diagnostics from the unpaid tail. STATE.completed means bounded delivery ended normally, including a budget-limited receipt; inspect budgetLimited and deliveredCount before assuming every expected variant was delivered. On interruption or a failed charge acknowledgement, delivery is incomplete and persisted dataset rows might remain.

API and CLI callers must set the maximum charge explicitly for each run. Include the maxTotalChargeUsd query parameter in the Run Actor API request, for example maxTotalChargeUsd=0.006. Stored Actor defaults do not guarantee the effective limit for every launch method. Check the returned run options for the requested value and isMaxTotalChargeUsdSetByUser: true, then inspect the final receipt. A partially delivered, budget-limited receipt does not confirm the remaining inventory.

See the Pricing tab for the configured platform price. Under this PPE contract, platform usage is included in the row price. Local core use has no service charge. No external model API, proxy, subscription or additional payment provider is needed.

The source package supports local CLI use as well as Actor execution. Synthetic test results do not establish customer acceptance or a live Webflow integration.

Fixture provenance and support

fixtures/PROVENANCE.txt distinguishes public official response shapes/IDs from fabricated relationships and intentional missing variants. These are synthetic engineering tests, not customer screenshots or a live reproduction of a vendor incident. Review the exact JSON/CSV contract before supplying data. Use the Issues tab after publication for reproducible, redacted input failures; never post tokens or private content.