Actor Trust Check: Score Any Apify Actor Before You Call It
Pricing
$2.00 / 1,000 actor checks
Actor Trust Check: Score Any Apify Actor Before You Call It
Is this Apify Actor reliable and safe to call? Get a 0-100 trust, quality and reliability score, badge and plain reasons from public stats: success rate, users, freshness, docs, pricing.
Pricing
$2.00 / 1,000 actor checks
Rating
0.0
(0)
Developer
Tudor Lane
Maintained by CommunityActor stats
0
Bookmarked
2
Total users
1
Monthly active users
11 hours ago
Last modified
Categories
Features
Share
Before your AI agent (or you) spends money on an Apify Actor, check whether it's likely to work. Pass an Actor name; get a 0-100 trust score, a badge and plain-English reasons, all from public data you can verify yourself.
Use when:
-
choosing between several Actors for the same job (e.g. "best Actor for Google Maps", "compare Instagram scrapers")
-
checking whether an Actor is reliable or safe to use before running it
-
verifying an Actor's success rate, freshness and pricing before spending money on it
-
Works on any public Apify Actor:
username/actor-name,username~actor-name, an Actor ID or an apify.com URL -
Score 0-100 and a badge: trusted (85+), ok (70-84), caution (55-69), avoid (under 55)
-
Plain reasons ("96.2% of 270,242 runs succeeded in the last 30 days", "Default build is 23 days old")
-
Every raw signal returned, so an agent can apply its own rules
-
Confidence (low/medium/high) so a new Actor with a handful of runs isn't confused with a proven one
-
Optional probe run (off by default, cost-capped at $0.01) that actually starts the Actor once
-
No text from the scored Actor's README or description is copied into the output, so nothing in it can steer your agent
Quick start
Click Start with the default input, or send:
{"requests": ["apify/web-scraper", "compass/crawler-google-places"]}
Send many Actors in one run (up to 100). Settings outside requests (probe, max_probe_usd) apply to every item.
Sample output
Real run, 20 Nov 2026 (default build 0.1.1):
| Actor | Score | Badge | Confidence | 30-day success | Users (30 days) | Build age | Pricing |
|---|---|---|---|---|---|---|---|
| apify/instagram-scraper | 100 | trusted | high | 99.7% | 71,219 | 0 days | pay per event |
| clockworks/tiktok-scraper | 100 | trusted | high | 98.4% | 38,027 | 3 days | pay per event |
| apify/website-content-crawler | 99 | trusted | high | 97.1% | 12,064 | 3 days | free |
| compass/crawler-google-places | 99 | trusted | high | 97.2% | 49,147 | 4 days | pay per event |
| apify/web-scraper | 97 | trusted | high | 96.2% | 1,364 | 23 days | free |
| tudorlane/sanctions-watchlist | 80 | ok | low | 100.0% | 1 | 22 days | pay per event |
| tudorlane/uk-property-sold-prices | 65 | caution | low | n/a | 1 | 0 days | pay per event |
Example reasons for apify/web-scraper:
["96.2% of 270,242 finished runs succeeded in the last 30 days (7,522 failed, 2,805 timed out; aborted runs not counted).","Default build is 23 days old.","1,364 users in the last 30 days, 127,333 all time, 182,063,998 runs all time; live 2,815 days.","Rated 4.7/5 from 100 reviews.","Docs: has input schema, output schema and a full README.","Pricing: free (you pay Apify platform usage)."]
The two tudorlane Actors are ours and are new, so they show low confidence; one has no public 30-day run stats yet.
How the score works
Rules version 2026-11-20.v1. The same rules apply to every Actor, including our own (tudorlane Actors are marked with a disclosure line). We don't change the rules after seeing results.
| Part | Points | What we measure (public Apify API and Store data) |
|---|---|---|
| Reliability | 40 | 30-day success rate = succeeded / (succeeded + failed + timed out). Aborted runs are left out (usually stopped by the user). Small samples are pulled towards a neutral 85% (worth 10 runs). 98%+ scores 40, 70% or lower scores 0. |
| Freshness | 20 | Age of the default build: up to 30 days 20, 90 days 16, 180 days 12, 365 days 6, older 0. A failed default build scores 0. |
| Adoption and track record | 20 | Users in the last 30 days (up to 8), total runs (up to 4), days live (up to 3), Store reviews (up to 5; fewer than 3 reviews counts as neutral 2). |
| Documentation | 10 | Input schema 3, README of 1,500+ characters 4 (300+ gets 2), output/dataset schema 2, description 1. |
| Pricing clarity | 10 | Free, pay-per-event or pay-per-result with named events 10; monthly rental 6; pricing not visible 3. |
Caps: an Apify notice (such as under maintenance) caps the score at 40; deprecated caps it at 20; no default build caps it at 30. A probe that fails with no results takes 10 points off.
Confidence is low under 50 finished runs or 5 users in 30 days, medium under 1,000 runs, otherwise high.
Probe run (optional)
Set "probe": true to also start one short test run of each Actor, in your own Apify account, using the Actor's own example input with a 60-second timeout. It only runs when the cost can be capped under max_probe_usd (default $0.01, max $0.05):
- Pay-per-event (the developer pays the compute, so memory doesn't affect your cost): Apify's hard spend limit (
maxTotalChargeUsd) is set to your cap. Skipped if the Actor demands a higher minimum. - Pay-per-result: limited to 1 result, and only if one result costs less than the cap.
- Free: run at the Actor's own default memory, estimated at memory x 60 s of platform usage; skipped if over the cap or the default memory is above 1 GB (probing with less memory than the developer intended would be unfair).
- Monthly rental or unknown pricing: always skipped. The output says whether it ran, why it was skipped, the status, result count, run time and usage.
Use cases
- An AI agent checks a tool before calling it and picks the most reliable of several similar Actors
- Developers vet a third-party Actor before building it into a pipeline
- Schedule a weekly check of the Actors you depend on and catch drops in success rate early
- Actor builders see how their own listing looks on objective signals
Pricing
Pay per event: $0.002 per Actor check. Rows with errors (e.g. Actor not found or not public) are not charged. No subscription. Probe runs, if you turn them on, are billed by Apify to your account at the scored Actor's own prices (capped as above).
Output
One dataset row per Actor: actor, actor_id, title, url, score, badge, confidence, reasons, score_breakdown, flags, pricing, signals (all raw numbers used), probe, disclosure, rules_version, checked_at. Errors come back as a row with error set, so a batch never fails because of one bad item.
Use from AI agents (MCP)
Add it to any MCP client via https://mcp.apify.com?actors=tudorlane/actor-trust-check, or call the Apify API / run-sync endpoint.
FAQ
Is this an official Apify rating? No. It's an independent score from Apify's public data. Apify's own Store badges are shown in signals.store_badge when present.
Why is a brand-new Actor marked caution? With few runs and users there's little evidence either way. The score leans neutral and confidence says low; check again after a few weeks.
Do you score your own Actors more kindly? No. Same rules, and the output says when the Actor is ours.
Can it check private Actors? No, only public ones.
Honest limits
- Public stats cover all users' runs, including runs that failed because of bad input, so a powerful Actor used in tricky ways can look less reliable than it is.
- Speed and data quality aren't in the public stats; the optional probe gives one run's speed only.
- A probe uses the Actor's example input; it can fail for reasons that aren't the Actor's fault (e.g. it needs your own API key).
- Ratings with fewer than 3 reviews are ignored. Store data can lag by a few hours.