Secret Scanner
Detect secrets exposed in a page or its assets: AWS, Stripe, GitHub, and OpenAI keys, plus generic high-entropy strings that look like credentials.
Critical33High15Medium3Low6Info1
Critical33 checks
- Stripe webhook signing secret in client bundleStripe whsec_* values must live server-side. If they show up in client code, attackers can forge webhook events.
- Google OAuth client_secret in sourceGoogle client_secret values must never appear in client-side code.
- AWS Secret Access Key in sourceAWS secret access keys (40-char base64) must never be in client code or public buckets.
- GitHub PAT in sourceghp_* / gho_* / ghu_* / ghs_* / ghr_* tokens grant repo / user access.
- NPM auth token in sourcenpm_ tokens allow publishing packages as you. Should never be in client code or public repos.
- PyPI token in sourcepypi-AgEIcHlwaS... tokens allow uploading packages.
- Algolia admin API key in sourceAlgolia admin keys (long base64 strings) grant full search-index control.
- Supabase service_role JWT in sourceThe service_role JWT bypasses Row Level Security. Should NEVER be in client code.
- PEM private key in source-----BEGIN ... PRIVATE KEY----- blocks grant signing/decryption access.
- Oracle Cloud Infrastructure (OCI) credentials in sourceOracle config files (.oci/config), user OCIDs, tenancy OCIDs, and API signing keys (PEM) grant full tenancy access.
- IBM Cloud IAM API key in sourceIBM IAM API keys (32+ char strings) grant access to the entire IBM Cloud account - Kubernetes, databases, Object Storage.
- DigitalOcean PAT in sourcedop_v1_* tokens allow full control over droplets, databases, spaces, and Kubernetes clusters.
- Linode API token in sourceLinode personal access tokens (64-char hex) grant Linode account takeover including instance termination.
- Vultr API key in sourceVultr API keys grant instance, DNS, and billing control on the Vultr account.
- RubyGems API key in sourceRubyGems API keys (rubygems_* / older 32-hex) allow publishing gems as the user, enabling supply-chain takeover.
- NuGet API key in sourceNuGet API keys (oy2_*) allow pushing new package versions to nuget.org under the user's identity - direct supply-chain attack.
- JFrog Artifactory API key in sourceArtifactory access tokens and encrypted passwords grant upload, delete, and admin over artifact repositories.
- GitLab runner registration token in sourceRunner registration tokens allow anyone to attach a malicious runner and exfiltrate CI secrets and source from every job.
- Bitbucket app password in sourceBitbucket app passwords (with associated username) grant repo, project, and account API access. Easy to misuse because they look like strings.
- PayPal OAuth client secret in sourcePayPal client secrets are paired with a client ID and grant order/refund/payout operations on the merchant account.
- Braintree API token in sourceBraintree access tokens (production + sandbox) grant full payment processing, refund, and customer data access.
- Square webhook signature key in sourceSquare HMAC signature keys let attackers forge webhook events (payments.created, refund.updated). Must live server-side only.
- Twilio API Key (SK prefix) in sourceTwilio API keys (SK*) with their secret allow programmatic SMS, voice, and account access - different from the Account SID (AC*) which is public-ish.
- MessageBird access key in sourceMessageBird access keys grant SMS, voice, WhatsApp, and contact-list access on the business account.
- Vonage / Nexmo API key + secret in sourceVonage (formerly Nexmo) API key/secret pairs grant SMS, voice, verify, and conversation API access.
- Meilisearch master key in sourceThe Meilisearch master key grants full control: index/document CRUD, tenant creation, and key issuance.
- Typesense admin API key in sourceTypesense admin keys grant full schema, document, and key-management control on the cluster.
- PlanetScale database password in sourcePlanetScale service-token JWTs and DB passwords grant full MySQL access including branch databases and deploy-requests.
- Auth0 client secret in sourceAuth0 client secrets paired with client IDs let attackers exchange authorization codes for tokens against any Auth0 application.
- Okta API token (SSWS) in sourceOkta SSWS tokens grant full admin access to the org - user CRUD, app assignments, factor reset. Treat like a root password.
- Keycloak realm signing key in sourceKeycloak realm RSA/EC private keys sign every JWT issued by the realm. A leaked key lets attackers mint valid tokens for any user.
- Cloudflare R2 access key in sourceCloudflare R2 secret access keys, paired with an r2.cloudflarestorage.com S3-compatible endpoint, grant full read/write access to R2 buckets.
- Resend API key in sourceResend API keys (re_*) grant the ability to send email from your verified sending domains and read email/audience data.
High15 checks
- Cloudflare API key in sourceCloudflare API tokens look like 40-char hex strings and grant DNS / WAF / Workers access.
- Tailscale auth key in sourcetskey-* values allow joining your tailnet. Rotate immediately if exposed.
- Mapbox secret token in sourcesk.* Mapbox tokens grant upload / dataset access. Use pk.* for client.
- PagerDuty REST API key in sourcePD keys grant incident / on-call rotation control.
- Datadog API key in sourceDatadog API keys grant metric ingestion. Restrict by hostname tag.
- HuggingFace write token in sourcehf_* tokens grant model / dataset upload access.
- Pinecone API key in sourcePinecone keys grant vector-DB control.
- Honeycomb write key in sourceHoneycomb events API keys grant ingest to specific datasets; long-lived classic keys can also read.
- GitLab deploy token in sourceGitLab deploy tokens grant push/pull to the project registry and package registry. Long-lived and project-scoped - very useful to attackers.
- Replicate API token in sourceReplicate API tokens (r8_*) grant inference and model upload access; can be abused for huge GPU bills.
- Cohere API key in sourceCohere trial/production keys grant access to generate, embed, and classify endpoints with billable usage.
- Mistral AI API key in sourceMistral API keys grant chat/completion/embedding access on the La Plateforme and can incur significant cost.
- Groq API key in sourceGroq API keys (gsk_*) grant high-throughput inference on Groq LPU hardware - attractive to attackers because of speed.
- Perplexity API key in sourcePerplexity API keys (pplx-*) grant billable access to Perplexity's chat completion / search endpoints.
- Linear API key in sourceLinear personal API keys (lin_api_*) grant full read/write access to a Linear workspace's issues, projects, and comments via the GraphQL API.
Medium3 checks
- Google Maps API key in sourceGoogle Maps API keys (AIzaSy*) in client code can be abused to bill against your account.
- AWS Access Key ID in sourceAKIA* keys are not secrets by themselves but pair with secret keys.
- High-entropy value assigned to a secret-shaped variable in sourceA response includes a long, high-entropy string literal assigned to a variable or JSON key named like a token, API key, or auth key: a shape common to real credentials that don't match any known vendor token format yet.
Low6 checks
- Firebase API key (public) in sourceFirebase Web API keys (AIzaSy*) are public by design. Security depends on Firestore rules / App Check.
- Twilio Account SID in sourceTwilio Account SIDs (AC*) are not strictly secret but should not be public.
- New Relic browser key in sourceNew Relic browser license keys (NRBR-*) are shipped in client JS to report RUM data. They are scoped by allowlist but still leak account info.
- Datadog client token in sourceDatadog client tokens (pub_*) are intentionally shipped in browser bundles for RUM. They are read-only / ingest-only but leak account.
- Sentry DSN in sourceSentry DSNs (public key + org + project embedded in an ingest URL) are meant to ship in client bundles. The risk is event-quota abuse and project/org disclosure, not data access.
- PostHog project API key in sourcePostHog project API keys (phc_*) are write-only capture keys meant to ship in client bundles. The risk is event-quota/data-poisoning abuse, not read access to analytics data.
About the Secrets checks
- What does the Secrets category check?
- Detect secrets exposed in a page or its assets: AWS, Stripe, GitHub, and OpenAI keys, plus generic high-entropy strings that look like credentials. AWS keys, Stripe, GitHub, OpenAI, generic high-entropy strings.
- How many secrets checks are there?
- VulnRadar runs 58 checks in this category, ranging across critical, high, medium, low, info severity.
- Do I need to install anything to run the secrets checks?
- No. Paste a URL in the browser or call the REST API, and the scan itself runs from our servers against that URL. There is no agent and no appliance to deploy alongside your application. A browser extension exists if you want to scan the page you are looking at, but nothing has to be installed to use this.
Run the secrets checks on your site
VulnRadar runs all 58 of these against a URL in seconds, with the evidence and fix for every finding. No agent to install, just paste a URL.