Barracuda - an unretired vendor cloud key used in an ARTEX-style intrusion
Barracuda (Arve Kjoelen, CISO; blog Nov 11, 2026, after a Trust Center advisory on Nov 10) describes a cloud intrusion that used a valid access key the company had given a third-party AI vendor and never…
What happened
Barracuda (Arve Kjoelen, CISO; blog Nov 11, 2026, after a Trust Center advisory on Nov 10) describes a cloud intrusion that used a valid access key the company had given a third-party AI vendor and never retired. In October the vendor's development environment was compromised. The Trust Center says the key was stored in a configuration file in that vendor's GitHub repository; other Barracuda credentials in the same file had already expired or were later revoked. First use was Nov 1, 2026 at 04:33 UTC. Within minutes the caller mapped what the key could reach and read secret stores, including AWS access keys and Azure service principal secrets. It decrypted infrastructure-as-code state files and pulled credentials from instance metadata services and Kubernetes service accounts. Over the next day each secret opened the next system: cloud accounts, then a data analytics system, then a source-code access token, then Kubernetes clusters. On Nov 2, in under four hours, short-lived branches were pushed to a handful of repositories so a build workflow printed the build system's secrets into the job log. The secrets were encoded so the log's usual masking would not hide them. Where a shell was obtained, commands searched for environment variables named WALLET, MNEMONIC, SEED, and PRIV, Ethereum keystore files, extended private keys, seed-phrase files, and Kubernetes secrets with wallet-related names. Barracuda says no wallet material was on those systems. For persistence, a container mounted one node's filesystem read-only and harvested that node's credentials. An attempt to create a new administrative cloud identity was blocked. About 10 cloud accounts were touched in roughly 44 hours. Last activity was Nov 2 at 23:31 UTC. An engineer noticed unusual activity on Nov 4, the same day the vendor reported its own incident, and the container was removed that evening Pacific time. Resources were labeled with the letter i followed by digits and no human-readable words. Barracuda says those labels, the probe names, the order of operations, and the built-in credential and wallet routines match ARTEX (https://github.com/RuoJi6/ARTEX). Because many groups use ARTEX, the match describes the method, not a named actor. Barracuda found no direct evidence that Barracuda Managed XDR Cloud or Monitored Endpoint integration credentials were viewed, but logging was not detailed enough to rule it out, and those customers were asked to rotate them. ATT&CK coverage in the write-up is T1552.005 (cloud instance metadata), T1078.004 (valid cloud accounts), T1610 (deploy container), and T1611 (escape to host via the mounted node filesystem). FBI FLASH-20261113-001, CSA 261113 (https://www.ic3.gov/CSA/2026/261113.pdf), is a general ARTEX overview and does not name Barracuda.
Barracuda matches resource names, probe names, order of operations, and built-in credential and wallet hunting to ARTEX. The tool was not recovered, and the match does not identify the operator.
How it happened
Read the flow as text & sources
A catalog reconstruction of the supplied source account. Unknown steps remain explicit; an edge describes the reported relationship between steps.
- 01 · Unretired vendor key
A cloud access key given to an AI vendor was never retired. After the vendor's October compromise, the key was in a configuration file in the vendor's GitHub repository.
Hunt: Repeated credential attempts or a stolen session are followed by a successful login or resource access inconsistent with the owner.
Shared infrastructure and legitimate travel create noise. IoT, SharePoint and cloud identity require their own collectors and field mappings.
Source for step 1 - 02 · Secrets open the next system
From Nov 1 04:33 UTC the key reads secret stores, infrastructure-as-code state, instance metadata, and Kubernetes service accounts, then reaches analytics, a source-code token, and clusters.
Hunt: An unexpected process or tool accesses credential stores and is followed by token use or outbound transfer inconsistent with the owner.
Defender DeviceFileEvents is not a general file-read audit trail. Collect suitable object-access, EDR or application logs; legitimate developer tools also read secrets.
Source for step 2 - 03 · ARTEX fingerprints, tool not recovered
Resources named with i plus digits, probe names, the order of operations, and wallet hunting match ARTEX. Barracuda did not recover the tool and does not name the operator.
Hunt: Untrusted content or a changed tool definition precedes a file read, command, or remote call outside the user's intended task.
Process logs cannot recover model intent. Remote connectors and in-process interpreters may produce no local child process. Prompt content alone is not a reliable signature.
Source for step 3 - 04 · CI logs and a host mount
On Nov 2, short-lived branches make a build workflow print encoded secrets into job logs. A read-only host filesystem mount harvests node credentials. A new cloud administrator identity is blocked.
Hunt: A container or Kubernetes pod is created with privileged execution or a host-directory mount that can expose the underlying node and enable container escape.
Infrastructure, storage, observability, and security workloads may legitimately require elevated access. Docker and Kubernetes use different audit schemas, and creation does not prove escape.
Source for step 4
Why it matters
Reported entry point: Valid cloud access key issued to a third-party AI vendor, left unretired, and later found in a configuration file in the vendor's GitHub repository after that vendor's development environment was compromised in October. Reported impact: On Nov 2, short-lived branches make a build workflow print encoded secrets into job logs. A read-only host filesystem mount harvests node credentials. A new cloud administrator identity is blocked; reported targets include Barracuda cloud environment, including a data analytics system, source-code access, Kubernetes clusters, and the system that stores Barracuda XDR customer integration credentials. Barracuda says it found no direct evidence those customer credentials were viewed; applicability to your setup has not been assessed, so verify affected versions, permissions, and configuration in the original report.
What you can do
Protection & prevention
Retire third-party cloud keys when the work ends, and prefer short-lived, narrowly scoped credentials. Do not leave one secret able to read the next store, the source-code token, and the cluster credentials. Alert on bulk secret reads, instance-metadata credential requests, Kubernetes secret lists, a new pod with a hostPath mount, and a new cloud administrator identity. Treat a short-lived branch whose build log prints secrets, especially encoded values that slip past log masking, as a credential-exposure event. Practice rotating every secret in reach, including customer integration keys, without waiting for proof they were read.
Catalog defensive guidance based on the reported behavior. Confirm scope and current mitigations in the original source.
Detection queries
Draft query starters grouped by the attack stages where they apply.
Start with the logs you collect. Each hunt identifies its required visibility, field mapping and limits. Queries are catalog-authored drafts using reusable behavior profiles; test with known activity and your normal baseline before alerting.
01Unretired vendor keyCredential / session use outside baselineCustom schema · Draft
A cloud access key given to an AI vendor was never retired. After the vendor's October compromise, the key was in a configuration file in the vendor's GitHub repository.
What to look for
Repeated credential attempts or a stolen session are followed by a successful login or resource access inconsistent with the owner.
The first calls used a still-valid cloud key issued to a vendor. The GitHub config file is where the key was found, not a commit on a Barracuda repository.
Visibility: Defender-side logs where collected
Required logs
- Identity / service authentication logs
- Device, source-IP and session context
- Resource access and token issuance audit
Coverage limit. Shared infrastructure and legitimate travel create noise. IoT, SharePoint and cloud identity require their own collectors and field mappings.
Custom-schema candidates, not out-of-box detectors. Map your application / cloud / platform logs to EventType, Actor, Asset, Action, Target and CorrelationId. KQL uses a custom OpenAttacksAudit_CL table; SPL uses openattacks:audit. EventType values are proposed normalization names, not native vendor event names. Correlation and environment-specific alert logic remain to be built.
Sigma
title: Hunt - Credential / session use outside baseline
id: 7d73d919-e785-4d40-ab0b-f12949adfaf5
status: experimental
description: Repeated credential attempts or a stolen session are followed by a
successful login or resource access inconsistent with the owner.
author: MagicSword OpenAttacks catalog
date: 2026-10-23
falsepositives:
- Authorized activity matching this broad hunt; baseline and investigate
before alerting.
level: low
logsource:
category: application
product: openattacks_audit
definition: Custom normalized audit schema; map EventType, Actor, Asset, Action,
Target and CorrelationId from the required source logs first.
detection:
selection:
EventType:
- identity.login
- identity.token.use
condition: selection
KQL
// SCHEMA TEMPLATE: create/map this custom table before use; not a built-in Sentinel table.
// OpenAttacksAudit_CL: TimeGenerated(datetime), EventType/Actor/Asset/Action/Target/CorrelationId(string).
// Selects candidates only. Apply the stage hypothesis and correlation steps before alerting.
OpenAttacksAudit_CL
| where TimeGenerated > ago(7d)
| where EventType in ("identity.login", "identity.token.use")
| project TimeGenerated, EventType, Actor, Asset, Action, Target, CorrelationId
| order by TimeGenerated descSPL
sourcetype="openattacks:audit"
EventType IN ("identity.login", "identity.token.use")
| table _time EventType Actor Asset Action Target CorrelationId
| sort - _timeInvestigation steps
- Compare failures, successes and session reuse by account and source; inspect access to new resources.
- Check device registration, MFA and credential changes; do not equate travel anomalies with confirmed compromise.
02Secrets open the next systemSensitive store access → credential useCustom schema · Draft
From Nov 1 04:33 UTC the key reads secret stores, infrastructure-as-code state, instance metadata, and Kubernetes service accounts, then reaches analytics, a source-code token, and clusters.
What to look for
An unexpected process or tool accesses credential stores and is followed by token use or outbound transfer inconsistent with the owner.
Secret-store reads, infrastructure-as-code state, and Kubernetes secret lists. Instance-metadata credential requests are a separate metadata hunt.
Visibility: Defender-side logs where collected
Required logs
- File-read / API access telemetry for relevant secret stores
- Process identity and command lines
- Identity, token and outbound-network audit logs
Coverage limit. Defender DeviceFileEvents is not a general file-read audit trail. Collect suitable object-access, EDR or application logs; legitimate developer tools also read secrets.
Custom-schema candidates, not out-of-box detectors. Map your application / cloud / platform logs to EventType, Actor, Asset, Action, Target and CorrelationId. KQL uses a custom OpenAttacksAudit_CL table; SPL uses openattacks:audit. EventType values are proposed normalization names, not native vendor event names. Correlation and environment-specific alert logic remain to be built.
Sigma
title: Hunt - Sensitive store access → credential use
id: 46c46c62-2d83-4198-a8bf-d6637094527b
status: experimental
description: An unexpected process or tool accesses credential stores and is
followed by token use or outbound transfer inconsistent with the owner.
author: MagicSword OpenAttacks catalog
date: 2026-10-23
falsepositives:
- Authorized activity matching this broad hunt; baseline and investigate
before alerting.
level: low
logsource:
category: application
product: openattacks_audit
definition: Custom normalized audit schema; map EventType, Actor, Asset, Action,
Target and CorrelationId from the required source logs first.
detection:
selection:
EventType:
- secret.read
- credential.store.read
condition: selection
KQL
// SCHEMA TEMPLATE: create/map this custom table before use; not a built-in Sentinel table.
// OpenAttacksAudit_CL: TimeGenerated(datetime), EventType/Actor/Asset/Action/Target/CorrelationId(string).
// Selects candidates only. Apply the stage hypothesis and correlation steps before alerting.
OpenAttacksAudit_CL
| where TimeGenerated > ago(7d)
| where EventType in ("secret.read", "credential.store.read")
| project TimeGenerated, EventType, Actor, Asset, Action, Target, CorrelationId
| order by TimeGenerated descSPL
sourcetype="openattacks:audit"
EventType IN ("secret.read", "credential.store.read")
| table _time EventType Actor Asset Action Target CorrelationId
| sort - _timeInvestigation steps
- Identify the actor, process and exact store touched, then pivot to token use from new devices or infrastructure.
- Correlate reads with egress and rotate affected credentials after confirming exposure.
03ARTEX fingerprints, tool not recoveredUnexpected agent tool useCustom schema · Draft
Resources named with i plus digits, probe names, the order of operations, and wallet hunting match ARTEX. Barracuda did not recover the tool and does not name the operator.
What to look for
Untrusted content or a changed tool definition precedes a file read, command, or remote call outside the user's intended task.
Barracuda did not recover the tool. Resource names and probe order are a hunt clue, not a tool-call audit record.
Visibility: Defender-side logs where collected
Required logs
- Agent tool-call audit records with session / tool / destination
- Tool-definition and approval changes
- Endpoint process, file-access and network evidence where available
Coverage limit. Process logs cannot recover model intent. Remote connectors and in-process interpreters may produce no local child process. Prompt content alone is not a reliable signature.
Custom-schema candidates, not out-of-box detectors. Map your application / cloud / platform logs to EventType, Actor, Asset, Action, Target and CorrelationId. KQL uses a custom OpenAttacksAudit_CL table; SPL uses openattacks:audit. EventType values are proposed normalization names, not native vendor event names. Correlation and environment-specific alert logic remain to be built.
Sigma
title: Hunt - Unexpected agent tool use
id: c709e384-0e11-479e-ab4d-3a06e2431e4b
status: experimental
description: Untrusted content or a changed tool definition precedes a file
read, command, or remote call outside the user's intended task.
author: MagicSword OpenAttacks catalog
date: 2026-10-23
falsepositives:
- Authorized activity matching this broad hunt; baseline and investigate
before alerting.
level: low
logsource:
category: application
product: openattacks_audit
definition: Custom normalized audit schema; map EventType, Actor, Asset, Action,
Target and CorrelationId from the required source logs first.
detection:
selection:
EventType:
- agent.tool.call
- agent.tool.definition.change
condition: selection
KQL
// SCHEMA TEMPLATE: create/map this custom table before use; not a built-in Sentinel table.
// OpenAttacksAudit_CL: TimeGenerated(datetime), EventType/Actor/Asset/Action/Target/CorrelationId(string).
// Selects candidates only. Apply the stage hypothesis and correlation steps before alerting.
OpenAttacksAudit_CL
| where TimeGenerated > ago(7d)
| where EventType in ("agent.tool.call", "agent.tool.definition.change")
| project TimeGenerated, EventType, Actor, Asset, Action, Target, CorrelationId
| order by TimeGenerated descSPL
sourcetype="openattacks:audit"
EventType IN ("agent.tool.call", "agent.tool.definition.change")
| table _time EventType Actor Asset Action Target CorrelationId
| sort - _timeInvestigation steps
- Join input provenance, tool descriptions and calls by session ID and compare the action to the approved task.
- Inspect child processes and destinations; record whether execution was in-process, local, or through a remote SaaS connector.
04CI logs and a host mountPrivileged container or pod with host accessNative logs · Draft
On Nov 2, short-lived branches make a build workflow print encoded secrets into job logs. A read-only host filesystem mount harvests node credentials. A new cloud administrator identity is blocked.
What to look for
A container or Kubernetes pod is created with privileged execution or a host-directory mount that can expose the underlying node and enable container escape.
The read-only host filesystem mount is the pod event. A short-lived branch that prints secrets into a build log is a CI event this pod query does not cover.
Visibility: Container orchestration and runtime audit logs where collected
Required logs
- Kubernetes API audit or Docker daemon events
- Container runtime process and mount telemetry
- Cloud control-plane identity and admission-controller decisions
Coverage limit. Infrastructure, storage, observability, and security workloads may legitimately require elevated access. Docker and Kubernetes use different audit schemas, and creation does not prove escape.
Kubernetes audit candidates for privileged pods or hostPath volumes. The KQL and SPL examples need the respective Kubernetes audit sources and local field extraction.
MCP reference matches: Privileged Container Deployed (SigmaHQ), Linux Suspicious Privileged Container Execution (Splunk Security Content). The query below is a catalog adaptation, not a verbatim upstream rule.
Sigma
title: Hunt - Privileged Kubernetes Pod or Host Mount
id: 5a3e19f4-3f1f-46e9-974f-4985b4692d30
status: experimental
description: Detects pod creation requests containing privileged containers or hostPath volumes.
author: MagicSword OpenAttacks catalog
date: 2026-11-06
references:
- https://github.com/SigmaHQ/sigma/blob/master/rules/application/kubernetes/audit/kubernetes_audit_privileged_pod_creation.yml
falsepositives:
- Approved infrastructure, storage, observability, and security workloads
level: high
logsource:
category: application
product: kubernetes
service: audit
detection:
selection_create:
verb: create
objectRef.resource: pods
selection_privileged:
requestObject.spec.containers.securityContext.privileged: true
selection_hostpath:
requestObject.spec.volumes.hostPath.path|startswith: /
condition: selection_create and 1 of selection_privileged, selection_hostpath
KQL
AzureDiagnostics
| where TimeGenerated > ago(7d)
| where Category in ("kube-audit", "kube-audit-admin")
| where log_s has '"verb":"create"' and log_s has '"resource":"pods"'
| where log_s has '"privileged":true' or log_s has '"hostPath"'
| project TimeGenerated, Resource, ResourceGroup, SubscriptionId, log_s
| order by TimeGenerated descSPL
sourcetype="kube:apiserver-audit" verb=create objectRef.resource=pods
(requestObject.spec.containers{}.securityContext.privileged=true
OR requestObject.spec.volumes{}.hostPath.path="*")
| table _time user.username sourceIPs{} objectRef.namespace objectRef.name
requestObject.spec.containers{}.image requestObject.spec.volumes{}.hostPath.path
| sort - _timeInvestigation steps
- Identify the creating principal, image digest, namespace, security context, host mounts, and admission result.
- Compare with approved workloads, then inspect node-level file access, processes, and credentials.
Technical detailsTooling, impact & MITRE ATLAS
Attack context
- AI's role
- AI as operator
- Offensive frameworks
- ARTEX
- Additional role
- AI credential theft
- Initial access
- Valid cloud access key issued to a third-party AI vendor, left unretired, and later found in a configuration file in the vendor's GitHub repository after that vendor's development environment was compromised in October
Tooling as reported
AI products, models and artifacts
- ARTEX (https://github.com/RuoJi6/ARTEX); the tool was not recoveredCommunity-maintained fork; the original repository is unavailable.
Other tooling
- Cloud secret stores, infrastructure-as-code state files, instance metadata, and Kubernetes service accounts
- Short-lived Git branches that triggered a build workflow and printed secrets into job logs
Targets and reported impact
- Targets / sectors
- Barracuda cloud environment, including a data analytics system, source-code access, Kubernetes clusters, and the system that stores Barracuda XDR customer integration credentials. Barracuda says it found no direct evidence those customer credentials were viewed.
- Reported scale / exposure
- About 10 cloud accounts in roughly 44 hours, from Nov 1 04:33 UTC to Nov 2 23:31 UTC. Detected Nov 4, after the activity had stopped.
- Target geography
- United States
Barracuda is the named victim. The blog does not state which cloud region the accounts ran in, and it does not name the vendor's country.
Reported attribution
Unnamed. Barracuda ties the labels, probe names, order of operations, and credential and wallet hunting to ARTEX, and says that match does not identify the operator. The wallet searches are the evidence for a financial motive. AI-service credentials were also in reach; Barracuda does not claim they were the goal.
Structured attribution
- Actor
- Unnamed
- Country
- Unknown
- Motivation
- financially-motivated
- Attributed by
- Barracuda (Arve Kjoelen, CISO)
- Attribution confidence
- low
Detection rules
- Kubernetes Pod Created with a Sensitive hostPath VolumeELASTICView rule
Selected from a security-detections-mcp search for hostPath. Kubernetes audit. A new pod whose volume mounts a sensitive path from the node. Matches the container Barracuda described mounting the node filesystem, including a read-only mount.
- Container With A hostPath Mount CreatedSIGMAView rule
Selected from a security-detections-mcp search for hostPath. Kubernetes audit create of a pod with any hostPath volume. Low severity because DaemonSets also use hostPath.
- Kubernetes Pod Exec Cloud Instance Metadata AccessELASTICView rule
Selected from a security-detections-mcp search for cloud instance metadata. Kubernetes audit of pod exec commands that reference instance-metadata endpoints. Covers credential reads from the node metadata service during an exec. It does not cover a metadata call made only by the kubelet or by a process that never appears in the exec command line.
- AWS Secrets Manager Rapid Secrets RetrievalELASTICView rule
Selected from a security-detections-mcp search for GetSecretValue. AWS CloudTrail. Twenty or more distinct GetSecretValue or BatchGetSecretValue calls by one identity in a short window. Barracuda described secret-store reads that included AWS access keys. This rule does not cover Azure service principal secret reads.
- Kubernetes Secrets EnumerationSIGMAView rule
Selected from a security-detections-mcp search for Kubernetes secrets. Kubernetes audit list on the secrets resource, which is how a caller inventories service-account tokens and wallet-named secrets.
- AWS IAM AdministratorAccess Policy Attached to UserELASTICView rule
Selected from a security-detections-mcp search for AdministratorAccess. AWS CloudTrail AttachUserPolicy of AdministratorAccess. Barracuda says a new administrative cloud identity was blocked and does not name the API. A denied call may not match a successful attach. There is no cited rule for the blocked attempt itself.
- CI log secret dump and i-plus-digits namesGUIDANCE
security-detections-mcp search for GitHub workflow returned GitHub Workflow File Creation or Modification, an endpoint hunt for .github/workflows files on a workstation. Barracuda describes short-lived branches that triggered an existing build workflow so the job log printed encoded secrets. That push is not a local workflow-file write, so the endpoint rule is not cited. Hunt the git host for short-lived branches and for job logs that contain encoded secret material. Also hunt resource names that are the letter i followed only by digits. No cited rule covers Azure service principal secret reads.
Activity timeline
- First seen
- 2026-11-01
- Last updated
- 2026-11-13
Updates
- 2026-10-01 — Barracuda
October, day not stated: the AI vendor's development environment is compromised. The Barracuda key is later found in a configuration file in the vendor's GitHub repository.
- 2026-11-01 — Barracuda
04:33 UTC first use of the key. Secret stores, infrastructure-as-code state, instance metadata, and Kubernetes service accounts are read, and the chain continues into analytics, source code, and clusters.
- 2026-11-02 — Barracuda
Short-lived branches make a build workflow print encoded secrets into job logs. A read-only host mount is deployed. A new cloud administrator identity is blocked. Last activity 23:31 UTC.
- 2026-11-04 — Barracuda
An engineer reports unusual account activity. The vendor shares its own incident the same day. The container is removed that evening Pacific time.
- 2026-11-10 — Barracuda Trust Center
Trust Center advisory on XDR integration credentials. Customers who may be affected are asked to revoke and recreate them.
- 2026-11-11 — Barracuda
Arve Kjoelen publishes the 44-hour narrative and the ARTEX fingerprint match.
- 2026-11-13 — FBI
FLASH-20261113-001 (CSA 261113) summarizes ARTEX activity. It does not name Barracuda.
Source defensive notes
Retire third-party cloud keys when the work ends, and prefer short-lived, narrowly scoped credentials. Do not leave one secret able to read the next store, the source-code token, and the cluster credentials. Alert on bulk secret reads, instance-metadata credential requests, Kubernetes secret lists, a new pod with a hostPath mount, and a new cloud administrator identity. Treat a short-lived branch whose build log prints secrets, especially encoded values that slip past log masking, as a credential-exposure event. Practice rotating every secret in reach, including customer integration keys, without waiting for proof they were read.
MITRE ATLAS
AI-specific catalog annotations based on the reported behavior; these are not MITRE case-study assignments.
The initial key was left in a vendor repository, and the intrusion read secret stores, infrastructure-as-code state, instance metadata, and Kubernetes service accounts. Source
The first calls used a still-valid cloud access key issued to the vendor. Source
Barracuda attributes the probe order and the built-in credential and wallet routines to ARTEX's planner and worker agents. The tool was not recovered. Source
Secrets were carried out by follow-on use of stolen keys and by a build log that printed encoded secret values. Source
Sources & research credit
5 sourcesResearch by Arve Kjoelen · Barracuda
- Forty-four hours: Anatomy of a machine-speed intrusion blog.barracuda.com
- Barracuda XDR integration credentials trust.barracuda.com
- Barracuda links cloud breach to ARTEX AI-powered cyberattack www.bleepingcomputer.com
- Barracuda hacked amid surge in ARTEX-powered cyberattacks www.securityweek.com
- FBI FLASH-20261113-001 (CSA 261113): Threat overview of ARTEX. Does not name Barracuda. www.ic3.gov
Times, the GitHub configuration-file detail, the wallet variable names, the read-only host mount, the blocked administrator identity, and the account count are from the Barracuda blog and Trust Center advisory. The community ARTEX fork is https://github.com/RuoJi6/ARTEX. FBI FLASH-20261113-001 does not name Barracuda; it is cited only so that gap is explicit. Link chainstack-intruder-reached-production-user-password-2026-11, datadog-artex-aws-cognito-2026-11, microsoft-artex-github-secrets-keyvault-2026-11, and eye-security-artex-gitops-2026-11.
Record & downloads
- Source record name
- Barracuda - an unretired vendor cloud key used in an ARTEX-style intrusion
- Entry ID
- barracuda-artex-cloud-key-intrusion-2026-11
- Catalog review
- November 17, 2026
- Category
- adversary-use-of-ai
- AI involvement confidence
- vendor-asserted