Cybersecurity

CNAPP RFP Template (Free): Requirements and Scoring

Paste-ready CNAPP requirements (CSPM, CIEM, CWPP, KSPM, image, IaC and secrets scanning, attack paths, CDR), vendor questions that separate platforms, and a weighted scoring rubric.

16 pages Microsoft Word Updated November 14, 2026
Download Template
CNAPP RFP Template (Free): Requirements and Scoring preview
Cloud-Native Application Protection Platform (CNAPP) RFP Template

Preview the first pages

See the structure before you download. The Word file has every section listed below, ready to fill in.

What's inside

The sections in the document, in order.

  1. How to run a CNAPP RFP
  2. CNAPP RFP requirements
  3. Questions to ask CNAPP vendors
  4. Scoring rubric
  5. Frequently asked questions
  6. Related RFP templates

Use this CNAPP RFP template when you're buying a cloud-native application protection platform to secure your cloud accounts, Kubernetes clusters, containers, virtual machines and serverless functions from code to runtime, or replacing separate CSPM, CWPP and CIEM tools with one platform.

Because a CNAPP combines these functions in one platform, the template lists each function separately and marks every requirement Must-have or Nice-to-have, so a vendor that is strong in only part of the platform can show that honestly.

It's written for the cloud security, platform engineering and procurement team running the evaluation, and it covers when to issue an RFI first, the requirements to send, the questions that separate vendors, and how to score the answers.

What’s inside this template

  • A step-by-step plan for running the evaluation, with an example schedule and advice on when a CNAPP RFI should come first
  • Requirements for cloud posture (CSPM), Kubernetes posture (KSPM), entitlements (CIEM), workload protection (CWPP), image, IaC and secrets scanning, attack path analysis and cloud detection and response, each marked Must-have or Nice-to-have
  • Optional requirements for data security posture (DSPM) and AI workload coverage, so you can scope the purchase
  • Questions that test agentless and agent-based coverage, how findings are ranked, what scanning adds to your cloud bill, and which parts of the platform are separate modules
  • A weighted scoring rubric with 1-to-5 definitions
  • An editable Word RFP with response codes, a pricing table and a timeline

Download the editable Microsoft Word version below. Every requirement, vendor question and scoring weight on this page is in the document, ready to tailor and send.

How to run a CNAPP RFP

A CNAPP evaluation involves cloud security, platform engineering, development and security operations teams, and the platform needs read access across your whole cloud estate, so much of the early work is agreeing on scope, permissions and who will act on findings. The example schedule in this template runs about 12 weeks from kickoff to contract award; allow more if your security team reviews each vendor's cloud permissions before anything is connected, or if the proof of value covers several cloud providers.

  1. Agree on scope and what the platform replaces (weeks 1-2). List the cloud providers, accounts, subscriptions and projects in scope, along with Kubernetes clusters, virtual machines, serverless functions, container registries, code repositories and CI/CD pipelines. Note which current tools the platform should replace and which it must work alongside, such as a CSPM tool, container scanners, workload agents or IaC scanners. Record the numbers vendors will price against: cloud accounts, virtual machines, Kubernetes nodes, serverless functions and, if code scanning is in scope, repositories or developers.
  2. Assemble the buying team. Include cloud security and security operations, who will handle runtime and detection alerts, the platform engineering team that will grant cloud access and run the sensors, development leads whose teams will receive findings, and your compliance lead, plus procurement and legal. Agree early who owns fixing findings, because a CNAPP creates work for teams outside security.
  3. Decide what must come from one platform, and whether you need an RFI (weeks 2-3). Decide which functions you want from a single platform and which you may keep or buy separately, and set Must-have and Nice-to-have priorities to match. If that's still open, or your long list has more than five vendors, issue a short RFI asking which functions each vendor builds itself, which are licensed separately, how agentless and agent-based coverage work, and how workloads are counted for pricing, and use the answers to build a shortlist.
  4. Tailor and issue the RFP (weeks 3-6). Delete requirements that don't apply, adjust priorities, and add your environment details to Section 1. Give vendors three weeks to respond, with a written question-and-answer window in the first week.
  5. Score the responses (weeks 6-7). Have each evaluator score independently against the rubric before comparing notes, and set aside any vendor that misses a Must-have. Check every 'yes' for whether it is included in the proposed edition or sold as an add-on, whether it works on every cloud provider you use, and whether it is generally available today or still in preview.
  6. Run a proof of value with two or three vendors (weeks 7-10). Connect each platform read-only to the same accounts, ideally one production-like and one non-production account, and install the runtime sensor in a test Kubernetes cluster. Run safe, scripted attack simulations in the test cluster and compare what each platform finds: whether its top-ranked issues are really exploitable, whether each finding names an owner and a fix, and how many duplicates it raises. Track the time to a first full inventory, what scanning adds to your cloud bill and the sensor's resource overhead, and call references of a similar size and cloud footprint.
  7. Negotiate and award (weeks 10-12). Use the pricing table to compare like for like, including how auto-scaling workloads are counted, add-on modules and the scanning costs you'll pay your cloud provider. Settle renewal caps, pricing for modules you may add later and exit assistance before you sign.

CNAPP RFP requirements

Each line is written to paste straight into your RFP. Must-have means a vendor that cannot meet it is out; Nice-to-have earns extra points. Change the priorities to fit your situation, and delete what doesn't apply.

Functional

ID Requirement Priority
F1 Agentless discovery and a continuously updated inventory of cloud resources across [list cloud providers] through read-only cloud APIs, covering compute, containers and Kubernetes services, serverless functions, storage, databases, networking and identities, with onboarding at the cloud organization level so new accounts, subscriptions and projects are added automatically. State how soon a new resource appears and how often posture is reassessed. Must-have
F2 Continuous detection of cloud misconfigurations (CSPM), with built-in checks mapped to the CIS Benchmarks published for our cloud providers and to [list frameworks, for example NIST SP 800-53, ISO/IEC 27001 and PCI DSS], custom policies we write, and compliance reports by framework, account and team that we can export as audit evidence. State how custom policies are written. Must-have
F3 Effective permissions analysis (CIEM) for human and non-human identities, including users, groups, roles, service accounts and workload identities, resolved across identity policies, resource policies, group membership, inherited permissions and cross-account or cross-project trust, so we can see what each identity can actually do. Must-have
F4 Identification of unused and excessive permissions from activity logs over a lookback period we set, with least-privilege policy recommendations we review before applying, plus findings for inactive identities, long-lived access keys, console users without MFA, privilege-escalation paths and external accounts with access. Must-have
F5 Agentless scanning of virtual machines and container hosts using disk snapshots, covering operating system packages, language libraries, malware, secrets and insecure configuration, with nothing installed on the workload. State how often workloads are rescanned and how soon a new workload is scanned. Must-have
F6 Agent-based runtime protection for Linux hosts, Kubernetes nodes and containers that detects threats as they happen, such as reverse shells, crypto-mining, privilege escalation, container escape attempts and execution of binaries that were not in the original image (drift). State Windows support. Must-have
F7 Response actions from the runtime sensor, such as ending a process, stopping or pausing a container, or isolating a workload from the network, taken by an analyst or automatically only for the rules we enable. Nice-to-have
F8 Scanning of serverless functions for vulnerable dependencies, embedded secrets and overly broad execution roles. State whether runtime protection for serverless functions is offered, for which runtimes, and how it is deployed. Must-have
F9 Kubernetes security posture management (KSPM) for managed and self-managed clusters: cluster and node configuration checked against the CIS Benchmarks for Kubernetes and, where published, for our managed Kubernetes services, risky workload settings such as privileged containers, host access and missing resource limits, role-based access control (RBAC) analysis and missing network policies. List supported distributions and managed services. Must-have
F10 Container image scanning in registries, in CI/CD pipelines and for running images, covering operating system packages, language libraries, malware, secrets and risky build settings such as running as root. List the supported registries and package ecosystems. Must-have
F11 Scanning of infrastructure-as-code in pull requests and pipelines for [list formats, for example Terraform, Kubernetes manifests, Helm charts and each cloud provider's template format], with results shown to developers in the pull request. State which policies are shared with runtime posture checks, so the same rule applies before and after deployment. Must-have
F12 Detection of exposed secrets, such as cloud access keys, tokens, passwords and private keys, in code repositories, infrastructure-as-code, container images, virtual machine disks and serverless function code, with secret values masked. State whether the platform can check if a secret it finds is still valid. Must-have
F13 A security graph linking resources, identities, network exposure, vulnerabilities, secrets and data, with attack path analysis that shows how an attacker could move from an exposed entry point to sensitive data or high-privilege roles, and a query interface for our own searches and custom policies. Must-have
F14 Prioritization of findings by internet exposure worked out from actual network paths (firewall rules, network access lists, load balancers and gateways), exploitability (such as listing in the CISA Known Exploited Vulnerabilities catalog and EPSS scores), effective permissions, data sensitivity where known, and the business criticality we assign, with the evidence behind each ranking visible to the analyst. Must-have
F15 Use of runtime data to show which vulnerable packages are actually loaded or in use in running workloads, so findings in unused packages can be ranked lower. Nice-to-have
F16 Cloud detection and response (CDR): near-real-time detection of threats from cloud audit logs, Kubernetes audit logs and workload runtime events, such as unusual credential use, privilege changes, disabling of logging and unusual data access, correlated into incidents with an investigation timeline and mapped to MITRE ATT&CK, including its Cloud and Containers matrices. State the typical delay from event to alert. Must-have
F17 Remediation guidance for every finding, with console steps, command-line commands or infrastructure-as-code fixes; assignment to the owning team, identified from cloud tags, account structure or repository mapping; risk acceptance with a named approver and an expiry date; and automatic closure when a rescan confirms the fix. Must-have
F18 Code-to-cloud tracing that links a running workload or cloud resource to the container image, infrastructure-as-code file, repository, commit and pipeline that produced it, so fixes can be made at the source. State whether fixes can be proposed to developers as pull requests. Nice-to-have
F19 Guardrails that can be set per policy to alert only or to block: CI/CD gates that fail a build on findings above a severity we set, and Kubernetes admission control that rejects workloads that break policy, with time-limited exceptions that need approval. Must-have
F20 Data security posture management (DSPM): discovery and classification of sensitive data in cloud storage and databases, used to raise the priority of findings that expose it. State whether classification reads data contents, where it runs and whether it is licensed separately. Nice-to-have
F21 AI workload coverage, sometimes called AI security posture management (AI-SPM): an inventory of the managed AI services, models, model endpoints and related resources in our cloud accounts, with posture checks such as public exposure of model endpoints, broad access to training data and excessive permissions for AI services and agents. State which AI services are supported today. Nice-to-have

Technical and architecture

ID Requirement Priority
T1 One platform with one inventory, one risk model and one console across all proposed functions. Identify any function that runs as a separate product, uses a separate console or data store, or is licensed as a separate module. Must-have
T2 Read-only access by default, through roles we create from vendor-provided templates and cross-account trust or short-lived credentials rather than long-lived access keys. List every permission requested, and identify any that allow reading disk snapshots or data contents. Must-have
T3 A documented agentless scanning architecture: whether snapshots are scanned in our cloud accounts or in the vendor's environment, what leaves our accounts, how long anything is kept, and an estimate for our environment of the compute, storage and data-transfer costs that scanning will add to our cloud bills, with controls to schedule or limit it. Must-have
T4 Published details for the runtime sensor: supported operating systems, kernel versions and Kubernetes distributions, deployment method, how it collects events (for example eBPF or a kernel module), CPU and memory overhead, and its behavior if it fails or loses contact with the service, which must never stop the workload. State support for self-managed clusters and virtual machines in our own data centers. Must-have
T5 A feature matrix for each cloud provider we use [list], identifying any function (such as CIEM, agentless scanning, runtime protection, serverless coverage or CDR) that is missing or more limited on that provider. Must-have
T6 Choice of the region where the vendor stores our inventory, findings and logs, to meet data-residency requirements. List the available regions, and state support for [government or sovereign cloud regions] if we use them. Must-have

Integration

ID Requirement Priority
I1 Single sign-on for administrators, analysts and developers through our identity provider via SAML 2.0 or OIDC, with user and group provisioning via SCIM. Must-have
I2 Integration with our source code management and CI/CD platforms [list] for infrastructure-as-code, image and secrets scanning in pull requests and pipelines, through native integrations, plugins or a command-line tool. Must-have
I3 Streaming of findings, runtime and CDR alerts, and administrator audit logs to our SIEM and SOAR platforms via API, webhook or cloud storage, in a documented schema. State whether OCSF is supported. Must-have
I4 Two-way integration with our ticketing (ITSM) platform, plus notifications to our chat tools, so findings create tickets routed to owning teams and ticket status flows back to the finding. Must-have
I5 A documented REST API covering the inventory, findings, policies, exceptions and graph queries, with published rate limits. State whether platform configuration can be managed as code. Must-have
I6 Generation and export of a software bill of materials (SBOM) for each container image and workload in SPDX or CycloneDX format, and import of SBOMs we produce in our own pipelines. Nice-to-have

Security and compliance

ID Requirement Priority
S1 A current SOC 2 Type II report and ISO/IEC 27001 certification covering the proposed service and any components we run, such as sensors, scanners and pipeline plugins, available under NDA. Must-have
S2 FedRAMP authorization at the required impact level, or equivalent regional certifications for public-sector use. Make this Must-have if you are a public-sector buyer. Nice-to-have
S3 Separation of read and write permissions: any write access needed for automated remediation or response actions is optional, granted through a separate role, limited to the actions we enable, and logged. Must-have
S4 Masking by default of secrets and sensitive values found during scanning, in the console, reports, exports and alerts. The vendor must not keep disk or data contents after a scan; list anything retained, such as file paths, package lists or masked snippets, and for how long. Must-have
S5 Administrator access through SSO with MFA, role-based permissions that can be limited by cloud account, project, cluster or business unit, and an audit log of every configuration change, policy exception and remediation or response action. Must-have
S6 Isolation of our tenant's data, encryption in transit and at rest for everything the vendor stores, and a documented key-management approach. State whether customer-managed keys are available for vendor-stored data. Must-have
S7 A published vulnerability disclosure process and committed timelines for fixing security flaws, including in runtime sensors, scanners and pipeline plugins we run, with direct notice when a fix needs action on our side. State how sensor and plugin releases are signed or otherwise verified. Must-have
S8 A data processing agreement, a current list of sub-processors, and a contractual commitment to notify us of security incidents affecting our data within a defined time. State the time. Must-have

Implementation and support

ID Requirement Priority
M1 A phased onboarding plan that connects posture and identity analysis across all our accounts first, then agentless workload scanning, then pipelines and runtime sensors, with the expected time to a first complete inventory and the effort needed from our teams. Must-have
M2 A tuning phase, agreed with us, that sets severities, suppression rules and ownership mapping before findings are routed to engineering teams. Must-have
M3 Identification of who delivers the implementation (vendor, partner or both) and the named team's experience with deployments of similar size and cloud footprint. Must-have
M4 A service-level agreement for availability of the platform, with service credits. State how availability is measured. Must-have
M5 24×7 support for critical issues, with response-time targets by severity. State your targets. Must-have
M6 Control over runtime sensor and pipeline plugin updates, such as staged rollout to a pilot group first, plus advance notice of changes to detection rules, risk scores, policies or the cloud permissions we must grant, release notes, a public status page and root-cause reports for incidents that affect us. Must-have
M7 Training for security analysts, platform engineers and developers, a named customer success contact, and periodic reviews of coverage and open risk. Nice-to-have

Commercial and pricing

ID Requirement Priority
C1 Pricing broken down by metric (per workload, per resource, per cloud account or per credit) and term, showing list price, discount and net price. Define exactly what counts toward the metric, such as a virtual machine, Kubernetes node, container, serverless function, container image, repository or developer. Must-have
C2 How short-lived and auto-scaling workloads are counted (for example peak, average or periodic sampling), and the terms that apply if usage exceeds the licensed quantity. Must-have
C3 Identification of every capability in the proposal that is licensed separately, such as runtime sensors, CDR, CIEM, code and pipeline scanning, DSPM, AI workload coverage and AI assistant features. Must-have
C4 Prices for adding workloads or modules mid-term, co-terminated with the main agreement. State how long proposal pricing is held for modules we don't buy at signing. Must-have
C5 A cap on price increases at renewal, covering the subscription and any add-ons. State the cap. Must-have
C6 A proof of value in our own cloud accounts at no charge, with the vendor's engineering support. Nice-to-have
C7 Exit terms covering export of the inventory, findings history, policies, exceptions and audit logs, instructions for removing sensors and access roles, and deletion of our data with written confirmation at the end of the contract. Must-have

Questions to ask CNAPP vendors

These questions are designed to separate vendors, not to collect brochure answers. Ask for evidence: a demo, a document or a reference.

  1. Which of the proposed functions (CSPM, CIEM, KSPM, agentless scanning, runtime protection, pipeline scanning, CDR, DSPM) share one inventory, one risk model and one console, and which are separate products, separately licensed modules or recent acquisitions? How do findings flow between them?
  2. Connect to [one test cloud account] and show every permission your platform requests. Which ones let it read disk snapshots or data contents, and what changes if we turn on automated remediation or response actions?
  3. Where are disk snapshots scanned: in our accounts or in your environment? List what leaves our accounts, how long it is kept, how soon a new virtual machine is scanned, and what compute, storage and data-transfer costs scanning adds to our cloud bill for an environment like ours [number of accounts and workloads].
  4. Take the highest-ranked finding in the proof of value and show the full attack path behind it: the entry point, the network exposure, the vulnerability or misconfiguration, the identity and permissions involved, and what an attacker could reach. Which of these factors raised or lowered its ranking?
  5. A critical vulnerability is published in a library we use. Show every image and running workload affected, which of them are reachable from the internet, whether the vulnerable package is loaded at runtime, and which repository, commit and team produced each image.
  6. We'll run safe, scripted attack simulations in a test Kubernetes cluster, such as a reverse shell, a crypto-mining binary, a binary that was not in the original image and a request for credentials from the cloud metadata service. Show what your sensor detects, how quickly, how each detection maps to MITRE ATT&CK, and which response actions are available.
  7. An access key is used from an unfamiliar location to create a new administrator and turn off logging. Show how this is detected from cloud audit logs, how it is linked to activity on the workloads involved, and the typical delay from event to alert.
  8. What CPU and memory overhead does your runtime sensor add on our node types [list], which kernel versions and Kubernetes distributions are supported, and what happens to the workload if the sensor crashes or can't reach your service?
  9. Show one policy enforced first as alert-only and then as a block, both in a CI/CD pipeline and in Kubernetes admission control. How does a developer request an exception, who approves it, and how does it expire?
  10. Pick one role or service account in the proof of value and show its effective permissions, including inherited and cross-account access, which of those permissions were used in the last [90] days, and the least-privilege policy you would recommend.
  11. How do you group and deduplicate findings so that one root cause, such as a base image used by many workloads or an IaC module used in many accounts, becomes one fix and one ticket rather than hundreds?
  12. For each cloud provider we use [list], which functions are less complete than on your most fully supported provider? Answer per function, not in general terms.
  13. If your proposal includes AI features, such as AI workload coverage or an assistant for investigation and remediation, show what is generally available today, what data the assistant sends and where it is processed, and whether it can be turned off.
  14. Which features in your proposal are generally available today, which are in preview and which are on the roadmap? Give dates for roadmap items.
  15. Provide a reference customer of similar size and cloud footprint that moved from separate cloud security tools to your platform. Which tools did they keep, and why?

Scoring rubric

Score each criterion from 1 to 5, multiply by its weight, and add the results. The weights below are a starting point; agree on your own before any proposals arrive so the scoring can't be bent around a favorite.

Criterion Weight What a strong response shows
Posture and entitlement coverage (CSPM, KSPM, CIEM) 15% Misconfigurations and excessive permissions found across every cloud provider you use, mapped to the benchmarks you report against, with effective permissions resolved correctly in the proof of value.
Workload, image and pipeline scanning 15% Agentless scanning of every workload type in scope, image, IaC and secrets scanning in your own pipelines, and code-to-cloud tracing back to the owning repository and team.
Risk prioritization and attack path analysis 15% A short, credible list of critical issues, attack paths backed by evidence, and rankings that use exposure, exploitability, permissions and data sensitivity.
Runtime protection and cloud detection and response 15% Scripted attacks detected in the proof of value with acceptable sensor overhead, cloud audit log detections linked to workload activity, and response actions you control.
Remediation, guardrails and developer workflow 10% Clear fix steps routed to owners, two-way ticketing, deduplicated findings, and alert-or-block guardrails with time-limited exceptions.
Architecture, data handling and operations 10% One inventory and console across functions, read-only access by default, a clear account of what leaves your accounts, a credible scanning cost estimate, and usable integrations.
Implementation and support 10% A phased onboarding plan, a tuning phase before findings reach engineering teams, an experienced named team, and clear SLAs and support targets.
Commercials and total cost 10% A clearly defined workload metric, fair counting of auto-scaling workloads, add-ons stated, a renewal cap and fair exit terms over a three-year view that includes scanning costs.
Total 100%

Scoring scale

  • 5 Exceeds: meets every Must-have and most Nice-to-haves, shown in a demo or proof of concept, with a matching reference customer.
  • 4 Strong: meets every Must-have and some Nice-to-haves, with clear evidence.
  • 3 Adequate: meets most Must-haves; gaps have a credible workaround or a dated roadmap commitment.
  • 2 Weak: misses one or more Must-haves, or the answer is vague.
  • 1 Poor: does not meet the requirement, or no answer.

Frequently asked questions

What should a CNAPP RFP include?

An enterprise CNAPP RFP should set out your scope (cloud providers, accounts, Kubernetes clusters, workloads, repositories and pipelines) and say which functions must come from one platform. It then needs requirements for cloud and Kubernetes posture, entitlements (CIEM), agentless scanning and runtime protection, image, IaC and secrets scanning, attack path analysis, cloud detection and response, remediation and guardrails, each marked Must-have or Nice-to-have; questions about permissions, scanning costs and which parts are separate modules; integration, security and support terms; a pricing table with a clearly defined workload metric; and the rubric you'll score with. This template includes all of them.

What is the difference between CNAPP, CSPM, CWPP and CIEM?

Cloud security posture management (CSPM) checks how cloud services are configured, against benchmarks and your own policies. A cloud workload protection platform (CWPP) protects the workloads themselves, such as virtual machines, containers and serverless functions, through vulnerability scanning and runtime protection. Cloud infrastructure entitlement management (CIEM) works out what each identity can actually do in your cloud accounts and finds permissions that are unused or too broad. A CNAPP combines these with Kubernetes posture, pipeline scanning and cloud detection and response, and links the findings in one risk model so you can see which issues combine into a real attack path.

Should we buy one CNAPP platform or separate cloud security tools?

One platform gives you one inventory, one console and one view of risk across posture, identities and workloads. Depth can differ from one function to another, though, so you may decide to keep a separate tool for one area, such as runtime protection or code scanning. Decide which functions must come from one platform and make those Must-have; the rest of this template is marked so that a vendor strong in only part of the platform can say so honestly and still be scored on its merits.

Does a CNAPP need agents, or is agentless scanning enough?

Agentless scanning reads cloud APIs and disk snapshots, so it covers every account and workload quickly with nothing to install, and it suits inventory, posture and vulnerability scanning. It looks at workloads at intervals, though, so it can't see process and network activity inside a running host or container as it happens. Runtime sensors fill that gap, at the cost of deploying and maintaining software on your nodes and hosts. This template asks for both as separate requirements (F5 and F6), so you can see which vendors do each well.

How long does a CNAPP RFP take?

The example schedule in this template runs about 12 weeks from kickoff to contract award, including three weeks for vendor responses and three to four weeks for a proof of value in your own cloud accounts. Allow more time if your security team must review each vendor's cloud permissions before anything is connected, or if the proof of value covers several cloud providers. Onboarding follows the award in phases, starting with posture across all accounts and adding pipeline scanning and runtime sensors later.

Related RFP templates

Download Template

Cloud security (CNAPP) vendors to consider

Aqua SecurityCrowdStrikeDatadogFortinetGoogle CloudMicrosoftOrca SecurityPalo Alto NetworksQualysRapid7SysdigTenableTrendAIUptycsUpwindWiz

Listed A to Z from vendors' own descriptions. Listings are free, unranked and not an endorsement. See all Cloud security (CNAPP) vendors →

More cybersecurity templates