Skip to main content
CloudArq
categorycspm
scopeaws
modeagentless
category guide · aws · cspm

Agentless AWS CSPM: posture, cost, and AI risk in one audit

Cloud Security Posture Management is the practice of continuously checking an AWS account for the misconfigurations that quietly create risk. CloudArq is an agentless, read-only AWS CSPM — one read-only IAM role, no agents — that folds security posture, compliance-control mapping, cost waste, and a dedicated AI-workload lens into a single audit, each finding paired with the exact remediation steps.

Updated 2026-07-20 · ~10 minute read

163
checks in one audit
7
compliance frameworks
5
posture domains
0
agents installed
definition

What CSPM actually is (one section)

CSPM — Cloud Security Posture Management — is the discipline of continuously inspecting a cloud account for misconfigurations and risky settings, then guiding you to fix them. Not vulnerabilities in your code, and not live intrusions, but the configuration mistakes that leave a door open: a storage bucket set to public, an IAM policy with a wildcard, a database left unencrypted, CloudTrail switched off in a region, a security group allowing the whole internet to reach a management port. Many cloud breaches trace back to a setting like one of these — not an exotic exploit, just a configuration nobody meant to leave the way it is.

A CSPM tool reads the state of your account, compares it against known-good baselines and framework controls, and surfaces the gaps. The good ones go further and tell you exactly how to close each gap. CloudArq is a CSPM in that mould — with the scope widened to include cost waste and the newest surface, AI workloads, in the same pass.

the approach

Why agentless matters

Some posture tools want an agent on every host, or a collector deployed into your account that runs continuously. CloudArq takes the opposite stance. It connects through a read-only IAM role secured with an ExternalId, created from a CloudFormation template, and reads configuration metadata and CloudWatch metrics through the AWS APIs. There is nothing to install, nothing running inside your workloads, and no software to keep patched.

That matters for two reasons. First, blast radius: a read-only role that can look but never touch is a far smaller thing to trust than an agent with a foothold in every instance. Second, time-to-value: you can point CloudArq at an account and get a full posture read from a single read-only pass, without a rollout. The trade-off is honest — an external, agentless audit sees the control-plane and configuration surface, which is where the overwhelming majority of posture risk lives; it does not sit inside a host watching process-level behaviour, and it is not trying to.

coverage

What the posture review covers — 6 domains

The full audit runs all 192 CloudArq checks. The posture side groups into the domains below — each one a cluster of specific, provable conditions read from your AWS metadata. The identifiers are real scanner check IDs; each describes a capability the read-only audit surfaces, never an observed behaviour.

Identity & access

Who can touch what — the blast radius of a leaked credential or an over-broad role.

4 example checks
no_mfa_rootiam_policy_wildcardold_access_keysroot_access_keys

Network exposure

What is reachable from the internet — open management ports and public data paths.

4 example checks
open_sshopen_rdppublic_s3exposed_self_hosted_inference_endpoint

Encryption

Data protected at rest and in transit, with key rotation on customer-managed keys.

4 example checks
unencrypted_ebsunencrypted_rdscloudfront_no_httpskms_key_rotation_disabled

Logging & threat detection

You cannot respond to what you cannot see — trails, config recording, flow logs, GuardDuty.

4 example checks
no_cloudtrailconfig_not_enabledvpc_flow_logs_disabledguardduty_disabled

Data protection & backup

Recoverability and tamper-resistance — versioning, backups, point-in-time recovery, object lock.

4 example checks
no_versioning_s3no_backups_rdsdynamodb_no_pitrs3_no_object_lock

AI-workload posture

The newest attack surface — Bedrock agent permissions, invocation logging, and log-sink exposure.

3 example checks
bedrock_invocation_logging_offagentic_blast_radiusinvocation_log_sink_exposure
one audit

Four lenses, one read-only pass

A classic CSPM stops at security posture. CloudArq runs three more lenses over the same account in the same audit, so you are not stitching four tools together to see one picture.

posture

Security posture

Misconfigurations and risky settings across identity, network, encryption, logging, and data protection — the core CSPM surface, each finding with the exact fix.

compliance

Compliance mapping

Findings mapped to the controls of eight compliance frameworks (tier-gated), so a posture gap reads as a specific control gap — for evidence, not a certification.

cost

Cost waste

Idle, oversized, and orphaned resources across compute, storage, networking, and databases — a per-resource estimate and remediation steps, no headline savings claim.

ai

AI-workload lens

A dedicated lens over Amazon Bedrock agents, AgentCore, RAG, guardrails, residency, and AI spend — control-plane metadata only, read-only, never touching agent data.

The AI-workload lens is covered in depth on the AI Workload and Bedrock security pages; the compliance mapping has a per-framework breakdown starting from the SOC 2 checklist and the Well-Architected review.

how it works

How it works

Read-only, no agents, no spend cap, never auto-fix

The same connection model as every CloudArq audit.

5 facts
  • connectA read-only IAM role with an ExternalId — created from a CloudFormation template. No keys to paste, no agents to install, nothing running inside your workloads.
  • read-onlyThe audit reads configuration metadata and CloudWatch metrics through the AWS APIs, and nothing more. It never stores your credentials, application data, database contents, or S3 objects.
  • encryptWhat CloudArq does keep — the finding metadata — is encrypted at rest with AES-256-GCM and hosted in the EU (Helsinki).
  • fixDetection plus the remediation steps — never auto-fix. Each finding shows the exact CLI command, console path, or Terraform block; you apply the change and stay in control.
  • no-capNo spend cap and nothing is ever stopped or throttled. CloudArq surfaces the risk and the fix; it never risks your workloads with a hard stop or an automatic action.
complementary

Where AWS-native tools fit

AWS ships its own building blocks, and they are legitimate and useful. AWS Security Hub aggregates and normalises findings from AWS security services; AWS Config records how your resources are configured over time; GuardDuty is a threat-detection service; and Trusted Advisor runs a set of cost, security, and service-limit checks. All of these are signals from inside your own account.

CloudArq is an independent, agentless external audit. It adds the exact remediation for each finding, a mapping to compliance-framework controls, cost waste, and an AI-workload lens — a different vantage point, not a replacement. Many teams run the native services for continuous in-account signal and CloudArq for the outside-in review and the fix. There is a longer, honest side-by-side on the CloudArq vs AWS-native security page, and a build-vs-buy breakdown on the scanner build-vs-buy page.

product · output

What the posture findings look like

audit · lens · postureillustrative example
open_ssh
Security group allows 0.0.0.0/0 on SSH
sg-0example · eu-north-1
fix →
public_s3
S3 bucket readable by anyone
arn:aws:s3:::example-assets
fix →
no_mfa_root
Root account has no MFA
account · root
fix →
unencrypted_rds
RDS instance not encrypted at rest
db-example-orders
fix →
bedrock_invocation_logging_off
Bedrock model-invocation logging disabled
bedrock · eu-north-1
fix →

Illustrative only — sample rows, not a real account. Each real finding groups under its posture domain and its mapped compliance controls, and ships with remediation steps.

faq

Frequently asked

01What is CSPM, and how is CloudArq’s approach different?
CSPM — Cloud Security Posture Management — is the discipline of continuously checking a cloud account for misconfigurations and risky settings: public storage, over-broad IAM, unencrypted data, disabled logging, exposed network paths. CloudArq is an agentless, read-only AWS CSPM that runs the whole posture review from a single read-only IAM role. What sets it apart is scope in one audit: security posture, a mapping of findings to compliance-framework controls, cost waste, and a dedicated AI-workload lens — each finding paired with the exact remediation steps.
02Is CloudArq really agentless? How does it connect?
Yes. There are no agents, sidecars, or daemons to install, and nothing to run inside your workloads. CloudArq connects through a read-only IAM role secured with an ExternalId — you create the role from a CloudFormation template, and the audit reads configuration metadata and CloudWatch metrics through the AWS APIs. It never stores your credentials, application data, database contents, or S3 objects.
03Does CloudArq change anything in my account, or fix issues automatically?
No. The audit is read-only end to end and never auto-fixes. Every finding ships with remediation steps — the exact CLI command, console path, or Terraform block — but you apply the change and stay in control. It is detection plus the fix, never an automatic action on your account and never a hard stop on your workloads.
04Does CloudArq replace AWS-native tools like Security Hub or GuardDuty?
They are complementary, and the honest answer is you can run both. AWS Security Hub aggregates and normalises findings from AWS security services, GuardDuty is a threat-detection service, and AWS Config records resource configuration — all signals from inside your own account. CloudArq is an independent, agentless external audit that adds exact remediation for each finding, a mapping to compliance-framework controls, cost waste, and an AI-workload lens. Many teams use the native services for continuous in-account signal and CloudArq for the outside-in review and the fix.
05How does the posture map to compliance frameworks?
CloudArq maps its automated checks to the controls of eight compliance frameworks (tier-gated — see the pricing page for which tier maps which subset). It shows posture against those controls; it does not certify or guarantee compliance, and CloudArq the business holds no certifications itself. A scanner evidences configuration; the attestation is still an auditor’s job.
06Does the CSPM cover AI and Bedrock workloads?
Yes. Alongside the classic posture and cost checks, the Max tier adds a dedicated AI-workload lens covering Amazon Bedrock agents, AgentCore, RAG knowledge bases, guardrails, data residency, and AI spend. It reads control-plane and configuration metadata only, is read-only, and never touches agent data or model traffic. See the AI workload and Bedrock security pages for the full picture.

Run the CSPM on your own account

Connect a read-only IAM role and CloudArq runs all 192 checks across 6 posture domains, the compliance mapping, cost, and the AI-workload lens — one agentless pass, each finding with the exact remediation steps and nothing changed on your account.

For the security story in depth, see the security overview; for the hands-on how-to guides, the resources hub.