AI Exposure Scanner: Public Secrets, Files & AI-Crawler Policy

See what your own public footprint exposes to AI systems and the open web, in one passive, scored report. Point it at domains and GitHub orgs you own: it looks for secrets committed to your public GitHub repos (always masked), well-known sensitive files left on your web server, which AI crawlers your robots.txt lets train on your content, contact data visible on your homepage, missing security headers and missing SPF/DMARC records. Every finding comes with a severity and a suggested fix, plus a 0–100 score and a short summary.

Pricing (pay per event, no start fee): $0.039 per domain scanned, $0.0039 per public repo scanned, $0.0019 per finding reported. Example: one domain plus a GitHub org with 15 public repos and 10 findings costs about $0.12.

Scan only what you own or are allowed to test. The authorizedToScan checkbox is required. Without it the run stops before sending a single request to your targets, and nothing is charged.

When to use this Actor (LLM / tool selection)

Use it when a person, agent or workflow needs to:

• Check whether API keys, tokens or passwords were committed to an organization's public GitHub repos, without ever seeing the full values
• Check whether a website serves .env, .git/config, backups, phpinfo or API docs it probably shouldn't
• See which AI bots (GPTBot, ClaudeBot, Google-Extended, CCBot, PerplexityBot and others) may train on or answer from a site, based on robots.txt, ai.txt, llms.txt and noai meta tags
• Get a quick security-header and SPF/DMARC check for a domain
• Re-run the same snapshot on a schedule to see whether fixes landed

Do not use it for: targets you are not authorized to scan, penetration testing, private repositories, gists, logged-in areas, or a compliance audit. It is a passive snapshot, not a certification.

What it checks

Check (checks value)	What it does	Example finding
secrets	Lists the org's public repos (newest-pushed first), picks up to 40 likely files per repo (config, env, CI, infra, source) and matches rules for AWS, GitHub, Slack, Google/GCP, Stripe, SendGrid, Twilio, OpenAI, Anthropic, npm, JWTs, private keys, database URLs and generic high-entropy assignments	AWS Access Key ID in public repo acme/api with evidence aws-access-key-id: AK********LE (len 20)
exposed_files	One GET per path from a short fixed list: /.env, /.env.backup, /.git/config, /.git/HEAD, /config.json, /wp-config.php.bak, /backup.zip, /backup.sql, /.DS_Store, /phpinfo.php, /server-status, /openapi.json, /.well-known/openapi.json, /swagger.json, /api-docs, plus a .map check for up to 8 scripts the homepage itself loads. Reported only when a 200 response also matches the file's content signature	/.git/config is publicly readable
ai_policy	robots.txt rules per AI crawler, ai.txt, llms.txt, TDM/noai meta tags; which bots may train vs only search/answer	13 AI training crawler(s) are allowed by robots.txt
data_exposure	Email addresses and phone numbers on the homepage (masked), presence of security.txt	1 email address(es) exposed on the homepage with evidence f***y@proton.me
headers	HSTS, CSP, X-Content-Type-Options, X-Frame-Options, Referrer-Policy on the homepage	4 security header(s) missing
email_auth	SPF and DMARC TXT records	No DMARC record

Secrets are masked at the source. Only the rule, file, line and a redacted preview (first and last two characters plus the length) are written. Full values never reach the dataset, the log or OUTPUT. Each secret finding carries a confidence of likely_valid, likely_test or unknown (from the file path, placeholder words and entropy), and identical values repeated in one repo are collapsed into one finding with a count.

Input

Field	Default	What it does
authorizedToScan	false	Required to scan. Confirms you own the targets or have explicit permission
domains		Your domains, e.g. example.com (up to 20)
githubOrgs		Your GitHub organization or user names; only public repos are read
githubToken		Optional, stored as a secret. Raises GitHub's API limit from 60 to 5,000 requests/hour. Sent only to api.github.com, never logged
checks	all	Any of secrets, exposed_files, ai_policy, data_exposure, headers, email_auth
maxRepos	50 (≤ 300)	Public repos scanned per org
includeForks / includeArchived	false	Also scan forks / archived repos
minSeverity	info	Findings below it are not written or charged (they still count in the score)
requestDelaySecs	2 (1–30)	Minimum gap between requests to the same website host

Example input

{
  "authorizedToScan": true,
  "domains": ["example.com"],
  "githubOrgs": ["your-org"],
  "minSeverity": "low"
}

Output

One dataset row per finding (view Findings), error rows for anything that could not be checked, and a final summary row. The same summary goes to the run's OUTPUT record.

Finding row (from a real run on our own GitHub org, 2026-10-18)

{
  "type": "finding", "category": "secrets", "kind": "aws-access-key-id", "severity": "low",
  "title": "AWS Access Key ID in public repo fairquery/fairquery-agent-trust-mcp",
  "detail": "A string matching AWS Access Key ID is committed at src/server.ts:64. It looks like a test/placeholder value. The value is shown redacted only. This exact value appears 3 times in the repo.",
  "location": "https://github.com/fairquery/fairquery-agent-trust-mcp/blob/main/src/server.ts#L64",
  "evidence": "aws-access-key-id: AK********LE (len 20)",
  "confidence": "likely_test",
  "recommendation": "Revoke/rotate the credential if real, remove it from the code, purge it from git history (e.g. git filter-repo), and load secrets from environment variables or a secrets manager instead."
}

(That value is AWS's documented example key in our own test code, so it is correctly flagged low and likely_test.)

Run summary (OUTPUT)

status, overall_score (0–100), grade (A–F), category_scores, findings_total, findings_charged, findings_by_severity, per-domain details (reachability, AI-crawler table per bot, SPF/DMARC), per-org repo counts, requests, events, event_charges_usd_at_list_price and duration_secs. The score starts at 100 and subtracts a weight per finding; likely_test secrets weigh a quarter.

Pricing (pay per event)

Event	Price
Domain scanned (all selected website and DNS checks for one domain)	$0.039
Public repo scanned	$0.0039
Finding reported (at or above minSeverity)	$0.0019

No Actor-start fee and no monthly plan. Free: runs without the authorization checkbox, error rows, the summary row and findings below minSeverity. Set a maximum cost per run in the run options and the Actor stops before exceeding it.

Example	Cost
1 domain, 6 findings	≈ $0.05
1 domain + 15 repos, 10 findings	≈ $0.12
1 domain + 50 repos, 25 findings	≈ $0.28

How it behaves (passive by design)

• No port scans, logins, exploits, fuzzing or brute forcing. Only GET requests to public URLs, public GitHub API calls and DNS lookups.
• robots.txt is respected for every website request, including the well-known-file checks; disallowed paths are skipped.
• Each well-known path is requested at most once per domain. The list above is complete.
• Honest user agent: FairqueryExposureScanner/1.0 (+https://getfairquery.com; fairquery@proton.me).
• Polite rate: one request at a time per host with a minimum gap; a 403 or 429 stops all further requests to that host for the run.
• Private and internal addresses are refused (SSRF protection with DNS pinning; ports 80 and 443 only).

Memory / runtime

512 MB is plenty. In our tests one domain took about 45 s, and a 13-repo org about 85 s without a token.

Limitations

• A clean report is not proof of safety. It covers only what was public and checked at scan time. It is not a penetration test, not an audit and not a compliance certification.
• Secret detection is heuristic. It can miss secrets in formats it has no rule for, in files outside the 40 candidates per repo, or in git history, and it can flag harmless strings. Verify every finding before acting.
• Public repos only: private repos, gists and (by default) forks and archived repos are not read.
• Only the homepage is checked for headers, contact data and noai tags; the site is not crawled.
• Without a githubToken, GitHub's 60 requests/hour limit can cut large org scans short (reported as an error row).

FAQ

Will it show me the leaked secret? No. You get the type, file, line and a masked preview. Open the file on GitHub to see it, then rotate it.

Can I scan a competitor or a client? Only with their explicit permission. The checkbox records your confirmation; you are responsible for having it.

Why does a test key lower my score? It is still a credential-shaped string in public code. Test/placeholder findings weigh a quarter of a real one, and you can raise minSeverity to hide them.

Is this a replacement for GitGuardian or a security platform? No. Those monitor continuously, including private code. This is a cheap, on-demand snapshot of your public footprint.

Fairquery family

Other Fairquery tools on the Apify Store. Each one returns structured rows with machine-readable error codes, and error rows are never charged:

• Email Validator for Agents (MX, Disposable, Typo): syntax, MX, disposable, role and typo checks without sending mail; $0.79 per 1,000 emails
• Website Contact & Company Enricher: domain in, public business emails, phones, socials and company details out; $3 per 1,000 domains
• Website Tech, SEO & llms.txt Check for Agents: tech stack, SEO score, AI-crawler rules and llms.txt per site; $1.50 per 1,000 sites
• Sanctions & Watchlist Screener (OFAC, EU, UK, UN): fuzzy name screening against official sanctions lists; $3 per 1,000 names + $0.007 per run
• LLM Data Guard: PII & Secret Redactor: redact PII and secrets before text reaches an LLM; $1.50 per 1,000 documents + $0.20 per 1,000 findings
• Dependency Vulnerability Check for Agents: packages and lockfiles against OSV.dev; $0.50 per 1,000 packages + $1.50 per 1,000 vulnerabilities
• EDGAR Company Facts & Filings for Agents: SEC filings and XBRL facts by ticker, CIK or name; from $1.50 per 1,000 companies
• Company Jobs Feed: Greenhouse, Lever & Ashby: live jobs from employers' own ATS boards with age and repost signals; $0.99 per 1,000 jobs + $0.0005 per board
• Competitor Change Tracker: Pricing, Terms, Careers & Changelog: watch competitors' public pricing, terms, careers and changelog pages for real changes with plain-English summaries and webhook alerts; $0.00049 per page checked + $0.0049 per change
• MCP server @fairquery/agent-trust-mcp: one MCP server for Claude Desktop, Cursor and VS Code (npx -y @fairquery/agent-trust-mcp). It has 7 free tools that need no token, plus paid tools that call the PII, email, sanctions and OSV Actors with your own Apify token.

All Fairquery Actors: https://apify.com/fairquery

Disclaimer

Fairquery is not affiliated with GitHub, OpenAI, Anthropic, Google, GitGuardian or any other company mentioned. Use the Actor only on targets you own or are authorized to scan.