Skip to content
Solvex Space

Cybersecurity · free, no signup

Cloud IAM Policy Analyzer

Paste an AWS IAM policy document and see what it actually permits. Wildcards get argued about in review; the finding this tool exists for does not, because privilege escalation lives in pairs of permissions that are unremarkable on their own. iam:PassRole permits nothing by itself — it only hands an existing role to a service. ec2:RunInstances permits nothing by itself — it only starts an instance. Together they are an instance the caller controls, running as the account’s most privileged role, with that role’s credentials readable from the metadata service. Fifteen documented paths are checked, each naming the exact permissions it needs and the route it takes. Action patterns are expanded as IAM wildcards first, so iam:* is recognised as granting iam:PassRole rather than missed as a different string. Deny statements are read and reported as narrowing, never as grants. Also covers Action * on Resource *, service wildcards and what each one includes, Allow with NotAction, and wildcard principals with and without a condition.

Authorized, defensive use only. Check assets you own or lawfully public data. This tool is educational and never provides another party's private data.

Paste an AWS IAM policy and see what it actually permits. Wildcards get argued about in review; privilege escalation does not, because it lives in pairs of permissions that are unremarkable on their own. iam:PassRole only hands an existing role to a service. ec2:RunInstances only starts an instance. Together they are an instance you control, running as the account's most privileged role, with that role's credentials readable from the metadata service. Fifteen documented paths are checked, each naming its exact permissions and its route.

Nothing is fetched, queried or simulated — no account is contacted and no ARN is resolved. This reads one document, and what a principal can really do is this policy combined with its other attached policies, permission boundaries, SCPs and any session policy. Deny statements are never read as grants.

An identity or resource policy document, as JSON.

What happens to what you enter

  • The analysis runs on Solvex servers. What is recorded afterwards is a SHA-256 hash of your input and a coarse summary of the outcome — never the input itself, and never a result you could be identified from.
  • Tools that read public sources (certificate logs, DNS, a public chain) contact those sources from our servers through a fetcher that refuses private and internal addresses.
  • An email address is stored only where a tool offers to send you a report and you choose to give one.

Limits: ten runs of this tool per minute from one address, and three signed reports per minute. Past that the tool says so and tells you when to try again.

When a free check is not enough

This tool reads what is public and reports what it can see. The Identity & Access Management Audit engagement is the authorized, specialist-led version: signed scope, findings proven by hand, a signed report, and a retest.

Explore Identity & Access Management Audit
Cloud IAM Policy Analyzer — Solvex Space