WordPress Plugin Release Change Monitor

Watch an explicit list of public WordPress.org plugin slugs for changes to published versions, minimum WordPress/PHP requirements, update timestamps and full changelog evidence. Each successful observation includes the exact selected source values, their missing/false/null status, response provenance and a comparison against a caller-supplied complete snapshot.

This Actor reads public directory metadata. It never accesses a WordPress installation, uses credentials, installs plugins or tests compatibility, security, upgrade safety or site behavior. Published minimum requirements are author-supplied metadata, not verified compatibility findings.

Quick start

Node.js 22 or newer is required.

npm ci --ignore-scripts
npm test
npm run validate:schema
npm run smoke

The smoke command performs two public requests for akismet and hello-dolly and writes qa/live-smoke.json. To run the real Apify SDK entrypoint locally:

mkdir -p storage/key_value_stores/default
cp examples/input.json storage/key_value_stores/default/INPUT.json
NODE_OPTIONS=--max-old-space-size=128 npm start

Results are written to the default dataset and the default key-value store's OUTPUT record. Each run uses only its own default storage; it does not open shared named stores or read another user's state.

Input

{
  "pluginSlugs": ["akismet", "hello-dolly"],
  "maxResults": 2,
  "maxRequests": 4,
  "maxRetries": 1,
  "maxSeconds": 45,
  "maxResponseBytes": 1000000,
  "maxBudgetUsd": 0.01
}

pluginSlugs is required: 1–25 distinct lowercase directory slugs consisting of letters, digits and internal hyphens. URLs, credentials, site inventories, wildcards and arbitrary endpoints are rejected. The watchlist is sorted before hashing, so reordering the same slugs preserves scope. Locale is fixed to en_US.

Bound	Default	Hard maximum
Plugin slugs / result rows	25	25
HTTP attempts including retries	40	100
Retries per plugin	2	3
Source scan time	90 seconds	300 seconds
One source response	1 MB	2 MB
Total response bytes	8 MB	8 MB
Raw changelog per plugin	64 KiB	64 KiB
Retained snapshot items	1 MB	1 MB
Total JSON input	2.5 MB	2.5 MB
Input event budget	$0.25	$20

Processing is sequential. An attempt has a deadline of at most 15 seconds within the overall scan deadline. Retries apply only to network errors, HTTP 429 and selected 5xx statuses. Retry-After is honored if it fits the remaining deadline. Redirects, 401/403 and other non-200 statuses stop the scan; no proxy or access workaround is used. This single-plugin action has no pagination: the explicit slug list is the complete requested scope. A result cap below its size creates an incomplete scan.

The deployment memory default is 256 MB, with a 256–1024 MB selectable range. See QA.md for measured local RSS and the limits of that measurement. The internal scan deadline and byte limits apply regardless of the selected platform timeout.

Output and comparisons

Each dataset row contains slug, evidence, provenance, pluginUrl, scopeKey, completeScan, change and interpretation. .actor/dataset_schema.json describes and validates the actual row fields.

evidence.version, minimumWordPress and minimumPHP keep {value, raw, present}. An absent API field, explicit null and explicit false remain distinct; value is null when no nonempty published string exists. No PHP or WordPress minimum is guessed. lastUpdated keeps the raw timestamp and reports whether the explicitly supported date format parsed successfully.

evidence.changelog.rawHtml keeps the complete selected source HTML. Its raw hash detects formatting changes as source changes. text is decoded plain text with script/style and embedded content excluded; textSha256 tracks this normalization. Nothing is shortened. Oversized changelogs fail the scan. Raw HTML is untrusted source text: do not execute it or render it as trusted HTML. The dataset table uses text fields and does not render this HTML.

provenance includes the exact official API query, observation time, HTTP status, complete response byte count and SHA-256. Response hashes may change because of unselected source fields; comparisons use selected evidence, not the whole-response hash. No author/contact fields are selected for output.

Always read OUTPUT.diagnostics.completeScan and deliveryComplete before using a snapshot. A pending-delivery record with no reusable snapshot is saved before dataset publication. Failed or short publication clears comparison and snapshot output. Individual row completeScan describes the source scan, while the OUTPUT record is authoritative for the entire delivered run.

On the first complete run, OUTPUT.comparison.status is baseline. Save OUTPUT.previousSnapshot and pass that exact object as previousSnapshot on the next run. Complete equal evidence produces unchanged; differences produce observed_metadata_change with changed field names, exact before/after evidence and both provenance records. All successful plugin observations produce rows, including unchanged observations.

Snapshots are isolated by the full watchlist, source, locale and parser version. Validation rejects incomplete, oversized, duplicated, forged/noncanonical or differently scoped state. Caller-supplied historical provenance cannot be authenticated independently; retain snapshots from your own trusted runs.

A missing plugin, failed request or partial scan yields no removals or change assertions and no reusable snapshot. If a previously known version, minimum requirement or changelog becomes unknown, the comparison is conservatively suppressed as known_evidence_became_unknown. This could be a valid source edit or a source/parser problem; review it and omit previousSnapshot to establish a deliberate fresh baseline. No “plugin removed” conclusion is inferred.

examples/output.json and examples/changed-output.json are deterministic synthetic fixture examples. Actual bounded live evidence is in qa/live-smoke.json and qa/sdk-live-OUTPUT.json when generated.

Approved listing pricing and controls

The user-approved listing price is $0.001 per plugin-observation plus $0.005 per automatic apify-actor-start unit. This is an approved custom start price; Apify's documented default is $0.00005. A start unit covers allocated memory up to 1 GB, then one unit per additional GB. This Actor never manually charges the synthetic start event.

One paid observation means one accessible metadata row, whether baseline, unchanged or changed. Errors do not create paid dataset rows. At the approved prices, 25 observations at the default memory allocation cost $0.030 in event charges; two observations cost $0.007. These are estimates, not account configuration or verified platform invoices. The approved listing rate reflects a bounded one-request metadata observation with retained evidence; it has not been benchmarked against live platform costs.

In PPE mode the runtime reads actual configured event prices, counts already charged events, reserves any pending automatic-start difference and applies the smaller of the platform and input event limits. Positive automatic default-dataset pricing is rejected to prevent double billing alongside the custom event. Free events and zero budgets have explicit handling. Work and pushes stop before another event would exceed a known cap. A custom event-count delta verifies publication; the SDK's aggregate chargedCount can be two for one row because of a free synthetic dataset event.

Outside PPE mode, the input budget uses the approved listing rates as a conservative work bound and does not create event charges. The input budget cannot reverse an automatic start fee charged before the program begins. Configure the platform run cap and pricing minimum before launch. Any separately configured platform usage charges are outside this event-only estimate. See PUBLISHING.md; no prices are configured by these files.

Source and limitations

The official WordPress plugins_api() reference documents the plugin_information action, slug argument, requirement fields and sections. Its current core source uses https://api.wordpress.org/plugins/info/1.2/. This Actor uses that fixed HTTPS endpoint and does not fetch arbitrary plugin or publisher websites.

An observation captures what that endpoint supplied at that time. It does not promise real-time source updates, historical completeness, plugin availability on a particular site, or tested compatibility. See NOTICE.md and QA.md. Cloud build and controlled Apify runs were verified on 2026-10-11 at 256 MB. Baseline and same-scope unchanged runs each delivered two observations with complete scans and delivery. A $0.005 platform event cap made no source requests and emitted no rows; an exact $0.006 cap delivered one observation. The platform recorded one automatic start plus the expected custom events, with no paid default-dataset event. Owner testing incurred platform usage rather than customer event invoices; customer payment settlement and every maximum-size workload remain unverified. Original QA.md records local build-stage evidence; these cloud checks extend that dated record.