agentleFS
apifyapify/awesome-skills/skills/apify-actor-idea-validator/SKILL.md

apify-actor-idea-validator

Skill299 starsChanged 10 days agoApache-2.0View on GitHub

Check whether a planned Apify Actor is worth building before writing code. The user describes the Actor; the skill measures demand (Google Ads search volume and CPC for "<site> scraper / api / mcp / data" keywords, plus the target site's Similarweb traffic), checks for an official API, counts competing Actors on the Apify Store, then digs into competitors in proportion to how crowded the niche is: input-schema powerup coverage and pricing, issue-queue and run-failure quality audit, and n8n template demand. Ends with a Markdown report answering: is there demand, is there room against competition, and which value-added features would make the Actor stand out. Use when the user asks to "check the potential of an Actor idea", "validate an Actor idea", "should I build an Actor for <site>", "is there demand for a <site> scraper", "is it worth building a <site> API or MCP on Apify", or describes a planned Actor and asks whether it makes sense. Requires Apify CLI or Apify MCP server.

  • Installs packages

The file

Show its rules and text

We split this skill into the rules it gives an agent

15 rules. About 0 are common, 7 are distinctive, and 8 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 8

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Is there demand?: Search volume for the keywords people use to find this kind of tool, plus traffic on the target site.

    line 29
  2. Very specific:

    All reports go to ~/apify-reports/ (flat folder, created if missing), never into the working directory or the skill folder:

    line 61
  3. Very specific:

    If no site is named ("an Actor that finds podcast guests"), use the core noun phrase as brand and skip Step 4.

    line 87
  4. Very specific:

    Read the run's defaultDatasetId, then:

    line 104
  5. Very specific:

    Use rows with status: "delivered" (avgMonthlySearches, cpcUsd, competition, competitionIndex, lowTopOfPageBidUsd, highTopOfPageBidUsd, …

    line 112
  6. Very specific:

    Search for "<brand> api", "<brand> api pricing" and "<brand> developer api".

    line 140
  7. Distinctive:

    Keep only results on the site's own domain or developer portal.

    line 147
  8. Very specific:

    Read the official docs and pricing page (web fetch tool, or apify/rag-web-browser).

    line 148
  9. Distinctive:

    If the site hosts several distinct products (e.g.

    line 176
  10. Distinctive:

    Run the modules one after another and extract:

    line 189
  11. Distinctive:

    Run health: failure and abort rates.

    line 196
  12. Distinctive:

    If the user has already validated a related idea today (same domain), reuse its traffic, API and competitor data rather than paying for it …

    line 200
  13. Distinctive:

    Write the report using references/report-templates.md.

    line 212
  14. Very specific:

    Option A: Apify CLI (recommended): Every call carries --user-agent apify-awesome-skills/apify-actor-idea-validator, --json (or --format …

    line 238
  15. Very specific:

    Never put an Apify token in a URL query string.

    line 253

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Apify Actor Idea Validator

Turn a description of a planned Actor into a go / no-go recommendation backed by three answers:

  1. Is there demand? Search volume for the keywords people use to find this kind of tool, plus traffic on the target site.
  2. Is there room against competition? An official API check, how many Actors already cover it, how big they are, and how reliable they are.
  3. What should we build to stand out? Value-added features backed by evidence: uncovered powerups, recurring issues and run failures, n8n workflows, keyword intent.

The skill is standalone. The three research modules it calls live in references/ and are run inline, not as separate skills.

Example prompts

Prompts this skill handles:

  • "Is there potential for a Quora scraper?"
  • "I want to build an Actor that scrapes XING profiles, companies and jobs for DACH lead gen. Does it make sense?"
  • "Check the potential of an Actor that ranks the top writers in a Quora topic."

Out of scope (the boundary):

  • "Audit the issues on my existing Actor" or "scrape these 500 URLs for me". This skill evaluates an idea; it doesn't run production scrapes or review a single published Actor in depth.

Prerequisites

  • Apify account (sign up)
  • Authentication via one of:
  • Apify CLI v1.5.0+ (npm install -g apify-cli), or the Apify MCP server
  • curl for the public, unauthenticated Apify Store, issues and n8n endpoints

Typical cost per run: about USD 0.30–0.40 in Actor charges. That covers the keyword lookup (USD 0.19 fixed fee + about USD 0.01 per keyword with data), Similarweb (about USD 0.014) and, when run, the n8n template scraper (a few cents). Store, issue and run-health lookups are free public endpoints.

Output

All reports go to ~/apify-reports/ (flat folder, created if missing), never into the working directory or the skill folder:

FileWhen
apify-actor-idea-<slug>-<YYYY-MM-DD>.mdAlways: the verdict report
apify-market-research-<slug>.mdWhen market research runs
apify-actor-quality-audit-<slug>-<YYYY-MM-DD>.mdWhen the quality audit runs
<top-competitor-slug>-n8n-scout.mdWhen the n8n scout runs

Templates for all four are in references/report-templates.md. In chat, give the verdict, the three one-line answers and the paths. Don't paste the report.

Workflow

Step 1 — Understand the idea

The user must describe the planned Actor. If there's no description, ask for one in one line and stop. Extract:

FieldExample
domain: target site(s)zillow.com
brand: how people name the site in searchzillow
data_types: what is extractedlistings, agents, price history
actor_type: scraper / API wrapper / MCP server / enrichment / automationscraper
audiencereal-estate investors, proptech
market: country / language of the site's usersUS/en (default); DE/de for xing.com

If no site is named ("an Actor that finds podcast guests"), use the core noun phrase as brand and skip Step 4.

Step 2 — Build the keyword set

Start from <brand> scraper, <brand> api, <brand> mcp, <brand> data. Add 4–11 variants from references/keyword-patterns.md that fit data_types, actor_type and audience. Cap at 15 short phrases of 2–4 words (full questions return no volume). For a job-to-be-done idea (e.g. "top writers in a topic"), include a few use-case terms ("expert finder", "influencer discovery"), because demand often sits there rather than on the site name.

Show the list and the proposed market to the user and let them edit it before the paid call.

Step 3 — Demand: search volume

apify actors call steadyfetch/keyword-search-volume-scraper \
  --input '{"keywords": ["zillow scraper", "zillow api", "zillow mcp", "zillow data"], "country": "US", "language": "en", "mode": "metrics-only"}' \
  --user-agent apify-awesome-skills/apify-actor-idea-validator \
  --json 2>/dev/null

Read the run's defaultDatasetId, then:

apify datasets get-items DATASET_ID --format json \
  --user-agent apify-awesome-skills/apify-actor-idea-validator 2>/dev/null

Per row:

  • Use rows with status: "delivered" (avgMonthlySearches, cpcUsd, competition, competitionIndex, lowTopOfPageBidUsd, highTopOfPageBidUsd, trendDirection).
  • nothing_found counts as 0 volume. That's a real signal, not an error.
  • plan_limit_reached or try_again means "no data". Tell the user, and don't count it as 0.
  • Skip the _summary / run_summary row except to note cost.

Keyword difficulty = Google Ads competitionIndex + CPC. Label it paid-search competition in the report, not SEO difficulty. High CPC on <brand> api / <brand> data, or on use-case terms, means buyers with budget.

Compute total_volume and note the dominant intent bucket: scraper, API, MCP, data, or use case.

Step 4 — Demand: site traffic (run in parallel with Step 3)

apify actors call tri_angle/similarweb-scraper \
  --input '{"websites": ["zillow.com"]}' \
  --user-agent apify-awesome-skills/apify-actor-idea-validator \
  --json 2>/dev/null

From the dataset item, record totalVisits, globalRank, topCountries, trafficSources and snapshotDate. If there's no data (small site), record that and gate on search volume only.

Step 5 — Demand gate

Demand is promising if total_volume ≥ 500/month or monthly visits ≥ 10,000. If it's below both, show the numbers and ask whether to stop (verdict Don't build: weak demand, demand section only) or continue (the verdict must still mention the weak demand).

Step 5b — Official API check

An official API is a competitor in its own right.

  1. Search for "<brand> api", "<brand> api pricing" and "<brand> developer api". Use the agent's own web search tool if it has one; otherwise:
    apify actors call apify/google-search-scraper \
      --input '{"queries": "zillow api\nzillow api pricing\nzillow developer api", "maxPagesPerQuery": 1}' \
      --user-agent apify-awesome-skills/apify-actor-idea-validator \
      --json 2>/dev/null
  2. Keep only results on the site's own domain or developer portal. Ignore third-party scrapers, "unofficial API" wrappers and RapidAPI listings.
  3. Read the official docs and pricing page (web fetch tool, or apify/rag-web-browser).
  4. Classify api_status and act on it:
api_statusMeaningEffect
noneNo official data API (login / share plugins or ads APIs don't count)Normal buckets
free-unlimitedFree, no meaningful quota, covers the planned dataAsk: stop (Don't build: free official API) or continue
free-rate-limitedFree with quota or rate limitHalve the buckets
paidPaid plan or per-call feeHalve the buckets
partner-onlyApproval or partnership needed, or it covers only a small part of the planned dataHalve the buckets

Halved buckets: <1 / 1–2 / 3–5 / >5 instead of <2 / 2–5 / 6–10 / >10.

Step 6 — Count competing Actors

apify actors search "zillow" --json --limit 10 \
  --user-agent apify-awesome-skills/apify-actor-idea-validator 2>/dev/null

actors search caps at 10 results. When it returns 10, list the full set from the public Store API (no token needed):

curl -s "https://api.apify.com/v2/store?search=zillow&limit=100"

Dedupe on username/name. Keep only Actors that actually target the domain. Their title or name must mention the brand. Watch for substring false matches (e.g. "xing" also matches "Douyin Xingtu"). Store search also matches README text, so raw counts overstate competition.

If the site hosts several distinct products (e.g. XING jobs vs. members and companies), split the count by segment. A crowded jobs segment can sit next to an empty lead-gen segment.

Record per Actor: MAU (stats.totalUsers30Days), total users, rating, pricing and last modified. Compute competitor_count, leader share (top MAU ÷ total MAU), Apify-owned Actors (apify, clockworks, compass) and stale Actors (> 6 months).

Step 7 — Route by competition

Normal bucketsHalved bucketsRun
< 20Nothing extra. Good candidate; derive features from the description and keyword intent.
2–51–2Market research
6–103–5Market research + quality audit + n8n scout
> 10> 5Quality audit first, then ask whether to also run market research and/or the n8n scout

Run the modules one after another and extract:

  • Market research:
    • Powerups covered by 0–1 Actors (gaps), and high-MAU powerups covered by few Actors.
    • Pricing: normalise each competitor to USD per 1,000 results. If every priced competitor is above USD 10 per 1k, add "Undercut on price" as a differentiation item.
  • Quality audit:
    • Recurring bug clusters ("market pain points") and recurring feature requests ("potential powerups").
    • Run health: failure and abort rates. A high-MAU Actor with a high failure rate is an opening for a reliable alternative.
  • n8n scout:
    • The LLM steps users bolt onto scrapers (power-ups), high-view templates built on competitor scrapers (replica angles), and empty niches.

If the user has already validated a related idea today (same domain), reuse its traffic, API and competitor data rather than paying for it again. Say so in Methodology.

Step 8 — Verdict and report

  • Build: demand is promising and competitor_count is below the first bucket, or competitors are weak (stale, low-rated, buggy) with no clear leader.
  • Build with differentiation: demand is promising, competition exists, and there are concrete gaps (uncovered powerups, pain points, an unserved intent such as API or MCP with real volume, an empty segment).
  • Don't build: weak demand (user stopped at Step 5), a free unlimited official API (user stopped at Step 5b), or a crowded niche with a strong, well-maintained leader and no gap with evidence.

Differentiation ranking:

  • Rank value-added features above price. List "Undercut on price" (only when triggered) last, with this note: price differentiation is less effective than value-added differentiation. Competitors can match a price cut in minutes, while a missing capability takes them a build cycle.
  • Suggest a pricing unit that matches the value (per company lead, per ranked writer) when the value is in aggregation.

Write the report using references/report-templates.md.

Step 9 — Chat close-out

Verdict: …

  • Demand: one line
  • Official API: one line
  • Competition: one line
  • Differentiate with: top 2–3 features

Reports: absolute paths

Actor routing

User needActor IDTierBest for
Keyword search volume, CPC, Ads competitionsteadyfetch/keyword-search-volume-scrapercommunityStep 3 demand
Site traffic, rank, countries, sourcestri_angle/similarweb-scrapercommunityStep 4 demand
Official API discoveryapify/google-search-scraperapifyStep 5b (when the agent has no web search)
Read API docs / pricing pagesapify/rag-web-browserapifyStep 5b (when the agent has no web fetch)
n8n template library searchadvantageous_subcontra/n8n-template-scrapercommunityn8n scout module

Tier = apify (Apify-maintained, prefer) or community (third-party).

Calling Actors — choose your interface

Option A: Apify CLI (recommended). Every call carries --user-agent apify-awesome-skills/apify-actor-idea-validator, --json (or --format json for datasets get-items) and 2>/dev/null, as in the examples above. Fetch an input schema before building any input not shown here:

apify actors info "steadyfetch/keyword-search-volume-scraper" --input --json \
  --user-agent apify-awesome-skills/apify-actor-idea-validator 2>/dev/null

Option B: Apify MCP server (mcp.apify.com). Use search-actors, fetch-actor-details (parameter actor:, not actorId:), call-actor and get-dataset-items with the same inputs.

Public read-only endpoints used with plain curl (no token, no auth header):

  • Store listing: https://api.apify.com/v2/store?search=<kw>&limit=100
  • Actor stats: https://api.apify.com/v2/acts/<username>~<name>
  • Actor issues: https://apify.com/api/actor-issues/data?actorId=<id>&statusOrName=open|closed
  • n8n template details: https://api.n8n.io/api/templates/workflows/<id>

Never put an Apify token in a URL query string. If an authenticated REST call is ever needed, use the Authorization: Bearer header.

Troubleshooting

  • plan_limit_reached on the volume Actor: the free Apify plan allows one fresh keyword lookup per day. Continue on Similarweb data only, or rerun after 00:00 UTC. Rerunning keywords already fetched in the last 30 days is free.
  • All Actors show 0 issues: you read body.list instead of body.json.list. See references/quality-audit.md.
  • Store API returns stats for only some Actors: parallel bursts get throttled. Retry sequentially, or mark tiny Actors as "no stats".
  • n8n results full of unrelated templates: competitor-name keywords pull in off-topic hits. Apply the relevance filter in references/n8n-scout.md.
  • For cost guardrails and more recovery steps, see references/gotchas.md.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.