Plugin & Extension Marketplace Radar

See which plugins and extensions are rising, slipping, going stale, or shipping a new version — across WordPress.org, the VS Code Marketplace, and the Atlassian Marketplace, in one schema.

Built for indie makers benchmarking a competitor, agencies deciding what to replace, and anyone who wants a scheduled watch instead of a one-off scrape. Each run stores a snapshot. The next run compares against that history and emits signals. A failed fetch never overwrites the last good snapshot.

Unofficial, not affiliated with WordPress, Automattic, Microsoft, Visual Studio Code, or Atlassian.

What you get

• A flat listing row per plugin or extension, with rank, rating, reviews, installs, version, and a source URL.
• History fields (previousRating, rankDelta, reviewVelocityPerDay, and the rest) once a snapshot exists.
• Signals: rank-risers, rating-decliners, review-velocity, install-velocity, stale-updates, new-entrants, changelog-diffs.
• One digest row per run with counts and short highlights.
• Alert-only mode for scheduled webhooks: unchanged listings stay in the snapshot and out of the dataset.
• The first run is still useful. Rows are flagged isBaseline: true with current ratings, installs, and stale/new-entrant signals. Deltas are null until the next good run.

You're only charged for rows that pass validation.

Default run

The default input queries the WordPress.org tag seo and the VS Code Marketplace search markdown, 15 listings each. A live run of that input returned 30 listing rows plus a digest in about two seconds, with fill rate 1.0 and five real stale-update signals (VS Code extensions with no release in more than a year).

{
    "marketplaces": ["wordpress", "vscode"],
    "wordpressCategory": "seo",
    "vscodeSearch": "markdown",
    "maxItemsPerMarketplace": 15,
    "signalFilter": "all",
    "windowDays": 7,
    "staleMonths": 12,
    "alertOnlyOnChange": false,
    "emitWeeklyDigest": false,
    "snapshotName": "plugin-marketplace-radar"
}

An empty input {} uses those same defaults.

Example output

Baseline listing from the WordPress.org plugin API on 23 October 2026. History fields are null because this was the first snapshot. The full row is flat and includes every column in the dataset schema.

{
    "recordType": "listing",
    "status": "baseline",
    "coverage": "complete",
    "isBaseline": true,
    "marketplace": "wordpress",
    "listingId": "wordpress-seo",
    "name": "Yoast SEO – Advanced SEO with real-time guidance and built-in AI",
    "publisher": "Yoast",
    "version": "28.7",
    "rating": 4.8,
    "ratingRaw": 96,
    "ratingScale": "0-100",
    "reviewCount": 27821,
    "installCount": 10000000,
    "installCountSource": "active_installs",
    "downloadCount": 1013620826,
    "rank": 1,
    "rankScope": "query-result",
    "lastUpdated": "2026-10-13T05:27:00Z",
    "publishedAt": "2010-10-11T00:00:00Z",
    "daysSinceUpdate": 10,
    "stale": false,
    "signals": [],
    "signalsCsv": "",
    "sourceUrl": "https://wordpress.org/plugins/wordpress-seo/",
    "observedAt": "2026-10-23T14:08:17Z",
    "confidence": 0.93,
    "contentHash": "f132e9de133b32cdd74e3daf3780d354f05c976da9058d5c4c80241e2f303407"
}

The same default run flagged current-state stale extensions. This one had not shipped a release in 592 days:

{
    "recordType": "listing",
    "status": "baseline",
    "marketplace": "vscode",
    "listingId": "yzhang.markdown-all-in-one",
    "name": "Markdown All in One",
    "version": "3.6.3",
    "rating": 4.686,
    "reviewCount": 172,
    "installCount": 14814461,
    "installCountSource": "install_statistic",
    "daysSinceUpdate": 592,
    "stale": true,
    "signals": ["stale-updates"],
    "signalsCsv": "stale-updates",
    "signalBilling": "charged",
    "sourceUrl": "https://marketplace.visualstudio.com/items?itemName=yzhang.markdown-all-in-one"
}

The digest row (recordType: "digest") from that run: 30 listings observed, 30 emitted, 5 signals, fill rate 1.0, coverage complete, status baseline. Highlights included vscode · Markdown All in One · no release for 592 days.

A later run against a snapshot that still had an older version string for Yoast SEO produced a real changelog excerpt (HTML removed) and a version diff. The previous version below is whatever the snapshot had stored; the excerpt is the public 28.7 release notes:

{
    "isBaseline": false,
    "previousVersion": "0.0.1",
    "version": "28.7",
    "versionChanged": true,
    "versionDiff": "0.0.1 -> 28.7",
    "changelogExcerpt": "28.7 Release date: 2026-10-13 Yoast SEO 28.7 brings more enhancements and bugfixes. Find more information about our software releases and updates here. Enhancements Allows agents to update titles and descriptions via the respective Yoast Ability. …",
    "signals": ["changelog-diffs"],
    "signalBilling": "charged",
    "rankComparable": true,
    "previousRank": 1,
    "rankDelta": 0
}

A repeat of the same stale extension does not bill signal-detected again. The row keeps stale: true, activeConditionsCsv: "stale-updates", signals: [], and signalBilling: "already-reported".

If every marketplace fails, the dataset gets one unbilled status row instead of a fake success:

{
    "recordType": "status",
    "status": "blocked",
    "coverage": "blocked",
    "reason": "wordpress: the source returned HTTP 403",
    "marketplace": "all"
}

The key-value record OUTPUT repeats the counts: rowsCharged, rowsDropped, fillRate, chargedUnits, and error counters (proxy, rate-limit, blocked, network, other).

Pricing

This Actor uses pay-per-event pricing. There is no per-dataset-item charge. You are never charged for a failed or empty capture, a status row, or the same signal state twice.

Event	Price	When it fires
listing-snapshot	$0.005	One validated listing row written to the dataset
signal-detected	$0.02	One listing that has at least one newly detected signal
weekly-digest	$1.00	Once per run, only if emitWeeklyDigest is true
Start event	$0.005 per GB	Charged by the platform. This actor defaults to 512 MB

1,000 listing snapshots ≈ $5. 1,000 signal-detected events ≈ $20.

Run	Events	Event cost
Default check: 30 listings, 5 new stale signals, no digest	30 × $0.005 + 5 × $0.02	$0.25
Quiet daily rerun of those 30, no new signals	30 × $0.005	$0.15
Alert-only rerun with nothing new	digest row only, no listing charges	$0
Weekly watch: 200 listings, 25 new signals, digest on	200 × $0.005 + 25 × $0.02 + $1	$2.50

The run stops cleanly when the max total charge is reached. Signals that were detected but not billed are kept and retried next run, so a cap does not silently drop them. Listing rows that fail the schema or a 60% core-field fill rate are dropped and not billed (rowsDropped in OUTPUT).

Input

Field	Default	What it does
marketplaces	wordpress, vscode	wordpress, vscode, atlassian
wordpressCategory	seo	Browse preset (popular, new, updated, top-rated, featured) or a plugin tag. Prefix search: to force a keyword search
vscodeSearch	markdown	Search text. Results are ordered by install count
atlassianQuery	time tracking	Text search. Prefix category: to filter by category name
watchlist	empty	Always-included ids. Use a prefix when more than one marketplace is selected: wordpress:akismet, vscode:yzhang.markdown-all-in-one, atlassian:com.soldevelo.apps.worklogs
signalFilter	all	Limit which signals are emitted and billed. Metrics are still stored
windowDays	7	Rank, rating, review, and install deltas use the oldest snapshot inside this window
staleMonths	12	No public release in this many months counts as stale (30.44 days per month)
maxItemsPerMarketplace	15	Search page size, 1–100. Watchlist items are extra
alertOnlyOnChange	false	Skip listing rows with no new signal. The first run for a marketplace still returns every current row
emitWeeklyDigest	false	Charge one weekly-digest event. The digest row is written either way
snapshotName	plugin-marketplace-radar	Named key-value store for this watch. Use a different name for a different query

No API key is required. The actor calls public read endpoints only.

How a repeat run works

1. Fetch the official API for each selected marketplace.
2. Load the named snapshot store. Rank and rating deltas compare to the oldest snapshot still inside windowDays. Version changes compare to the most recent snapshot, so a release alerts once.
3. Write listing rows (unless alert-only mode suppresses them) and one digest row.
4. Save the new snapshot and the fingerprints of signals that were actually billed.

History keeps the last 40 successful snapshots. A blocked, empty, or crashed fetch does not replace the previous good one.

Signal rules, all visible on the row:

• Rank risers. Moved up at least 3 places in this query. Rank is the position in the fetched page, not a global store rank. Compared only when the query is unchanged.
• Rating decliners. Normalised 0–5 rating fell by at least 0.1. WordPress.org’s 0–100 score is divided by 20.
• Review velocity. At least one new rating since the comparison snapshot. reviewVelocityPerDay divides by the gap, with a one-hour minimum so a quick rerun does not explode the rate.
• Install velocity. Install count increased. WordPress.org active installs are coarse public buckets, so this fires when the bucket changes. Cumulative downloads are on the row as downloadCount and do not, by themselves, bill a signal.
• Stale updates. Latest public release is older than staleMonths. Available on the first run.
• New entrants. Published inside the window, or absent from the previous snapshot of the same query. A first run does not mark the whole result page as new.
• Changelog diffs. Version string changed. The new release notes are fetched as plain text (HTML stripped, length capped). Atlassian uses the vendor’s short release summary.

confidence is 0–1 based on how complete the official payload was. coverage is complete, partial, empty, blocked, or unsupported.

Scope and limits

In this version:

• WordPress.org plugin directory API (api.wordpress.org)
• Visual Studio Marketplace public extension query for VS Code
• Atlassian Marketplace REST API (/rest/2/addons)

Not in this version: Chrome Web Store, Shopify App Store, Slack, and Zapier. Those need a different access path and are intentionally out of scope.

Other limits:

• Rank is the position inside your query, capped by maxItemsPerMarketplace.
• WordPress.org active-install figures are rounded buckets. Use downloadDelta if you need a finer public series, and do not treat either as audited sales.
• Atlassian’s latest-version call is one extra request per app. Release notes are often a one-line summary; marketing HTML from the version page is not copied through.
• VS Code and WordPress.org changelogs are fetched only after a version change, and only from allowlisted HTTPS hosts.
• Reviewer names and review text are not collected. Counts and averages only.
• Descriptions are stored as a short plain-text snippet. Marketplace copy is untrusted text: tags are stripped and fields are length-capped.
• The watchlist accepts slugs and ids, not URLs. Private hosts, file://, and raw IP addresses are rejected.
• Same-day reruns compare correctly, but velocity over a few minutes is not a trend. Schedule daily or weekly.
• A crash after a charge and before the snapshot is saved can bill a signal again on the next run. The idempotency key covers retries inside a few minutes.

Integrations

Schedule the actor in the Apify Console (daily for a 7-day window, weekly if you only want the digest). Point a webhook at the default dataset.

• n8n / Make: run the actor, then keep items where recordType is listing and signalsCsv is not empty. The digest row is recordType digest.
• Google Sheets: download the dataset as CSV. signalsCsv, categoriesCsv, and highlightsCsv are single columns.
• Slack or email: use alert-only mode so a quiet day sends the digest and nothing else.
• Agents: every input has a default, nothing is both required and defaulted, and a run with no usable rows returns an explicit status row instead of an empty success.

Run locally

Python 3.12.

pip install -r requirements.txt
mkdir -p storage/key_value_stores/default
echo '{}' > storage/key_value_stores/default/INPUT.json
python -m src

Dataset items land in storage/datasets/default/. The run report is storage/key_value_stores/default/OUTPUT. Snapshots land in storage/key_value_stores/plugin-marketplace-radar/.

pip install -r requirements-dev.txt
pytest

Set ACTOR_TEST_PAY_PER_EVENT=true to log charge events locally. Nothing is billed off the platform. On the platform, create the three events above and leave the automatic dataset-item event off.

The actor asks for 512 MB and does not use standby mode.

Sources

Catalog metadata comes from the public WordPress.org plugin info API, the public Visual Studio Marketplace gallery query, and the Atlassian Marketplace REST API. Those providers keep their own terms. This actor sends a identifiable User-Agent, throttles requests, and retries with backoff.

Questions: plusonecares@gmail.com.