AI Tool Stack Detector

Tell us a company name — we'll tell you which AI dev tools their engineers actually use. Cursor? Claude Code? Copilot? Windsurf? Pulled from public signals: careers pages, eng blogs, GitHub orgs.

What does this Actor do?

Every modern engineering org is quietly rewiring around AI dev tools — Cursor, Claude Code, GitHub Copilot, Windsurf, Cody, Codeium, Aider, Continue, Tabnine, Cline. But there is no public registry of "who uses what."

This Actor reconstructs an org's AI tool stack from four public signals:

1. Careers pages: scrapes /careers, /jobs, follows job posts on Greenhouse / Lever / Ashby / Workable / Smartrecruiters. A line that says "Experience with Cursor or GitHub Copilot a plus" is high-confidence evidence.
2. Engineering blog: scrapes /blog, /engineering, follows recent posts. "We adopted Claude Code last quarter" is gold.
3. GitHub org: discovers the org via direct slug + search, then checks the top 20 most-recently-updated repos for filesystem markers (.cursorrules, CLAUDE.md, .windsurfrules, .github/copilot-instructions.md, .continuerc.json, etc).
4. Homepage + docs: scans the homepage, /docs, and /about for any tool name drops.

Every detection includes the source URL and the matching snippet. Confidence is weighted — explicit job-listing mentions beat repo-marker presence, which beats a generic homepage word match.

Why use AI Tool Stack Detector?

• AI dev-tool sales: build a target list of "Cursor users" / "Copilot users" / "no-AI orgs" without scraping LinkedIn
• Competitive intelligence: track which orgs are migrating from Copilot to Cursor in real time
• Engineering hiring: align your candidate pitch to the tools the target uses
• Investor research: identify the orgs that have actually adopted AI tools vs. announced they would
• Internal benchmarks: see what your competitors' eng teams are running

How to use AI Tool Stack Detector

1. Enter the company name (e.g. Vercel, Stripe, Anthropic)
2. (Optional) Provide homepageUrl to skip discovery
3. (Optional) Pick which signalsToCheck to run — default is all four
4. Click Start
5. Open the dataset row to see the full confidence-scored profile

Input

• company (string, required) — e.g. "Vercel"
• homepageUrl (string, optional) — skip homepage discovery
• signalsToCheck (array, default ["careers", "blog", "github-org", "docs"]) — subset to run

Example:

{
  "company": "Vercel",
  "homepageUrl": "https://vercel.com",
  "signalsToCheck": ["careers", "github-org"]
}

Output

{
  "company": "Vercel",
  "homepageUrl": "https://vercel.com",
  "githubOrg": "vercel",
  "aiToolsDetected": {
    "cursor": {
      "confidence": 0.9,
      "evidence": [
        "careers:https://vercel.com/careers/... → \"Cursor IDE\"",
        "github:vercel/next.js → \"Found .cursor/\""
      ],
      "signalCount": 5
    },
    "claude-code": {
      "confidence": 0.7,
      "evidence": ["github:vercel/turbo → \"Found CLAUDE.md\""],
      "signalCount": 2
    },
    "copilot": {
      "confidence": 0.5,
      "evidence": ["blog:.../post → \"GitHub Copilot\""],
      "signalCount": 1
    }
  },
  "primaryStack": "cursor",
  "signalsAnalyzed": 18,
  "timestamp": "2026-05-21T01:30:00Z"
}

Download as JSON, CSV, HTML, or Excel.

Data table

Field	Meaning
aiToolsDetected[tool].confidence	0.0 to 0.99 — weighted by source strength
aiToolsDetected[tool].evidence	Up to 5 evidence snippets with source URL
aiToolsDetected[tool].signalCount	Raw mention count
primaryStack	Tool with highest confidence
signalsAnalyzed	Total pages + repos scanned

Pricing

• First detection: free trial
• Per-company rate: $0.10 per company scanned
• Batch (50 companies): $4
• Monthly subscription: $10/month for unlimited scans + weekly refresh

Tips

• Set GITHUB_TOKEN env var to a personal access token — anonymous GitHub API is harshly rate-limited
• Run with signalsToCheck: ["careers"] for fastest scans (skips repo-by-repo GitHub checks)
• Schedule weekly and diff: catch the day a competitor switches from Copilot to Cursor
• Pair with gh-org-ai-fingerprinter for deep GitHub-only fingerprinting

FAQ

How accurate is the detection?

We never claim 100% certainty — we publish a confidence score (0.0-0.99) and the evidence. A 0.9 score means "multiple high-weight signals from multiple sources." A 0.4 score means "one ambiguous mention." Use confidence as a filter, not a binary.

What if a company has no public GitHub org?

The Actor falls back to careers + blog + homepage signals. No GitHub means lower max signal count, not failure.

Is this legal?

Yes — every signal is publicly indexed content (careers pages, blog posts, public GitHub repos). The Actor obeys robots-friendly pacing and never logs in. Do not use the output for adversarial outreach or anti-recruiting.

Why no Playwright?

We use axios only. Most careers + blog content renders server-side. Browser-based scraping is 30x slower and unnecessary for the 95%-case.

Support

Issues / feature requests: open in the Issues tab on the Apify console.

Built by Ian Mu — github.com/ianymu — author of claude-verify-before-stop, part of his 100-actor Apify portfolio.