Agentic AI in cybersecurity: pick a security job and see what its agents read, do alone and leave behind
Agentic AI in cybersecurity means agents that, in Anthropic's words, "dynamically direct their own processes and tool usage". On 19 Nov 2026 we counted 85 named agents from 39 vendors across six jobs; 14 of 28 triage vendors show no approval step. Test them with the evaluation guide, agent harness and benchmarks.
Every job below answers the same four questions about its agents: what an agent reads, what it does without a person, where a person approves, and what evidence it leaves. Each vendor quote is printed once, under the question it answers.
Vulnerability triage and remediation
An agent takes scanner findings, decides which matter in this environment, and moves them toward a fix.
Reads (What an agent reads)
4 of 28 vendors quoted on what the agent reads
Published agents read scanner findings and how the vulnerable component is used in the customer's code or environment (usage, reachability, exposure, ownership).
AikidoAikido Agent (CVE Exploitability Analysis)
“Instead of treating every CVE with the same urgency, the Agent reads how the vulnerable package is used in your code”
AsteliaThe Investigator (Automated Triage)
“Astelia agents triage each new vulnerability by reachability and operational impact, validate the finding, and coordinate next steps across your tech stack.”
Endor LabsAI SAST Remediation (AURI Agent)
“The agent reads it against your source at the pinned commit, works out what's real, and either writes the fix or files the exception.”
CrowdStrikeExposure Prioritization Agent (Charlotte AI, Falcon Exposure Management)
“The Exposure Prioritization Agent cuts through the noise by identifying what attackers are most likely to exploit and why”
Does alone (What it does without a person)Supplier’s settingsYours
17 of 28 vendors let customers build or change the agent
Vendors describe agents that rank or dismiss findings with a stated reason and, in some products, draft tickets or fixes. Which actions run unattended is a setting in several products, not a fixed behavior.
Cogent SecurityCogent AI agents (vulnerability lifecycle)
“Different agents handle normalization, enrichment, risk assessment, and remediation planning, each optimized for its step.”
MazeMaze remediation agents
“AI agents find the root cause, validate a fix, and route it to the responsible developer or coding agent.”
WizGreen Agent
“the Wiz Green Agent acts as a digital security investigator and remediation engine”
Tools and permissions
If you buy: Inside the supplier’s settings. 17 of 28 vendors listed for this job state on their pages that customers can build or change the agent; for the other 11, none found on 19 Nov 2026.
Source
NIST AI RMF 1.0: “MAP 4: Risks and benefits are mapped for all components of the AI system including third-party software and data.”
If you build: You grant each tool and each credential.
A person approves (Where a person approves)Where offeredYours
14 of 28 vendors describe an approval step
Vendors describe an approval step before changes: configurable per action, an approval queue, or review and merge of a pull request. The point where approval sits differs by product.
Aikido
“With Human Approval, the agent proposes the action and waits for you to approve it from the analysis history.”
Cogent Security
“Decide where agents act independently and where they pause for sign-off. Configure by action type, severity, or asset sensitivity.”
CrowdStrike
“Actions can be set to execute autonomously or require human approval before running”
Astelia
“The drafted ticket and the proposed firewall rule wait in the approval queue for a human click”
Wiz
“Route lower-confidence remediation verdicts from the Green Agent to a developer in Slack for approval”
Approval rules
If you buy: Where the product offers them. 14 of 28 vendors listed for this job describe an approval step on the pages read 19 Nov 2026.
If you build: You write which actions wait for a person.
Evidence (What evidence it leaves)Product recordsYours
17 of 28 vendors describe an audit trail
Vendors describe logs of agent steps and decisions, with reasons attached to each decision.
Aikido
“Finding is ignored with the reason CVE determined not exploitable”
Astelia
“every step lands in the execution log, and every decision in the audit log.”
Cogent Security
“every agent action is logged with its reasoning.”
CrowdStrike
“role-based permissions, execution traces, audit logs, credit caps, and configurable approvals.”
Wiz
“Teams can follow this analysis directly in the Investigation tab, where each step in the Green Agent’s analysis is laid out.”
Logs
If you buy: What the product records. 17 of 28 vendors listed for this job describe an audit trail on the pages read 19 Nov 2026.
If you build: You log every step where the agent cannot edit it.
Model choice
If you buy: The supplier picks and updates the model.
Source
NCSC, Guidelines for secure AI system development: “You support users to evaluate and respond to model changes (for example by providing preview access and versioned APIs).”
If you build: You pick the model and can swap it.
Source
Anthropic, Scaling Managed Agents: “This allows the implementation of each to be swapped without disturbing the others.”
Each row opens to the vendor’s own words from its own page. All read 19 Nov 2026; counts from the vendors directory.
All 28 vendors for this job, lane by lane
A to Z, from each vendor’s own pages read 19 Nov 2026. Dot: stated; ring: none found.
- AikidoAgentAikido Agent (CVE Exploitability Analysis)Build or changeCustomers can build or change itApproval stepApproval step describedAudit trailAudit trail described
- Amplify SecurityAgentAmplify Console triage and remediation agentsBuild or changeCustomers can build or change itApproval stepNone foundAudit trailNone found
- AsteliaAgentThe Investigator (Automated Triage); The Fixer (Agentic Remediation)Build or changeCustomers can build or change itApproval stepApproval step describedAudit trailAudit trail described
- AverlonAgentMitigation AgentsBuild or changeNone foundApproval stepNone foundAudit trailNone found
- BacklineAgentAndy (Analyst); Polly (Planner); Vera (Verifier)Build or changeNone foundApproval stepNone foundAudit trailNone found
- CheckmarxAgentTriage AssistBuild or changeNone foundApproval stepNone foundAudit trailAudit trail described
- Cogent SecurityAgentCogent AI agents (vulnerability lifecycle)Build or changeNone foundApproval stepApproval step describedAudit trailAudit trail described
- CorridorAgentCorridor AgentBuild or changeCustomers can build or change itApproval stepNone foundAudit trailNone found
- CrowdStrikeAgentExposure Prioritization Agent (Charlotte AI, Falcon Exposure Management)Build or changeCustomers can build or change itApproval stepApproval step describedAudit trailAudit trail described
- CycodeAgentExploitability AgentBuild or changeCustomers can build or change itApproval stepApproval step describedAudit trailAudit trail described
- Dux SecurityAgentDux AI agentsBuild or changeNone foundApproval stepNone foundAudit trailNone found
- Endor LabsAgentAI SAST Remediation (AURI Agent)Build or changeCustomers can build or change itApproval stepApproval step describedAudit trailNone found
- GitHubAgentAssign security campaign alerts to CopilotBuild or changeCustomers can build or change itApproval stepApproval step describedAudit trailAudit trail described
- MazeAgentMaze remediation agentsBuild or changeNone foundApproval stepApproval step describedAudit trailNone found
- MicrosoftAgentVulnerability Remediation Agent (Security Copilot in Intune, public preview)Build or changeCustomers can build or change itApproval stepNone foundAudit trailNone found
- Onit SecurityAgentExploitability Agent; Ownership Agent; Remediation Agent; Mitigation AgentBuild or changeNone foundApproval stepApproval step describedAudit trailAudit trail described
- PixeeAgentPixee triage engine (AI agents tier)Build or changeCustomers can build or change itApproval stepNone foundAudit trailAudit trail described
- ProjectDiscoveryAgentVuln Triage Agent (Neo); Verification Agent (Neo)Build or changeCustomers can build or change itApproval stepNone foundAudit trailAudit trail described
- QualysAgentAgent Val; Agent SaraBuild or changeNone foundApproval stepApproval step describedAudit trailAudit trail described
- Rival SecurityAgentTriage findingsBuild or changeCustomers can build or change itApproval stepNone foundAudit trailAudit trail described
- SeemplicityAgentHost VM AnalystBuild or changeNone foundApproval stepNone foundAudit trailAudit trail described
- SemgrepAgentSemgrep Multimodal (formerly Semgrep Assistant); Semgrep Agentic Workflows (review step)Build or changeCustomers can build or change itApproval stepNone foundAudit trailNone found
- SnykAgentEvo Remediation AgentBuild or changeNone foundApproval stepApproval step describedAudit trailAudit trail described
- TenableAgentTenable Hexa AIBuild or changeCustomers can build or change itApproval stepApproval step describedAudit trailNone found
- TorqAgentTorq agentic AI in vulnerability management (HyperAgents)Build or changeCustomers can build or change itApproval stepNone foundAudit trailAudit trail described
- VicariusAgentAgentic Exploit Simulation (vIntelligence)Build or changeNone foundApproval stepNone foundAudit trailNone found
- WizAgentGreen AgentBuild or changeCustomers can build or change itApproval stepApproval step describedAudit trailAudit trail described
- ZafranAgentAgentic RemediationBuild or changeCustomers can build or change itApproval stepApproval step describedAudit trailAudit trail described
None found for this job: approval step, 14 of 28 vendors (Amplify Security, Averlon, Backline, Checkmarx, Corridor, Dux Security, Microsoft, Pixee, ProjectDiscovery, Rival Security, Seemplicity, Semgrep, Torq, Vicarius); audit trail, 11 of 28 (Amplify Security, Averlon, Backline, Corridor, Dux Security, Endor Labs, Maze, Microsoft, Semgrep, Tenable, Vicarius); customer can build or change the agent, 11 of 28.
Response to a new disclosure
When a new CVE or advisory is published, an agent finds which of the customer's assets are affected and proposes what to do now.
Reads (What an agent reads)
4 of 10 vendors quoted on what the agent reads
Published agents read the new advisory against the customer's software inventory, SBOMs and asset data, sometimes before scanner signatures exist.
AsteliaThe Hunter (Zero-Day Detection)
“Astelia agents analyze newly disclosed and zero-day vulnerabilities against your SBOM and software to find what’s reachable, even before scanners detect them.”
Cogent SecurityCogent AI agents (new disclosure)
“Identify vulnerable software within minutes of a new disclosure, before scanners can.”
ZafranZero Day Agent
“Zafran's Zero Day Agent matches it to your live software inventory in 88 seconds”
SeemplicitySCA Analyst
“powers SBOM-driven zero-day response before scanner signatures even exist.”
Does alone (What it does without a person)Supplier’s settingsYours
4 of 10 vendors let customers build or change the agent
Vendors describe agents that match a disclosure to assets and hand back a short list of affected systems with next steps. Some also generate detection or mitigation scripts.
Onit SecurityZero-Day Agent
“The Zero-Day Agent picks up a new critical vulnerability as it breaks, runs it against your environment, and hands your team a short set of decisions with the evidence attached, in minutes.”
QualysAgent Insta
“Agent Insta decides what to correlate next based on new advisories, live inventory, and threat context”
VicariusScriptAI
“When a vulnerability is being actively exploited with no fix available, ScriptAI generates detection and remediation scripts so your team is not left waiting on the vendor.”
MazeMaze Cloud agents
“Triage and remediate CVEs in containers and VMs, and catch zero-days before scanners see them.”
Tools and permissions
If you buy: Inside the supplier’s settings. 4 of 10 vendors listed for this job state on their pages that customers can build or change the agent; for the other 6, none found on 19 Nov 2026.
Source
NIST AI RMF 1.0: “MAP 4: Risks and benefits are mapped for all components of the AI system including third-party software and data.”
If you build: You grant each tool and each credential.
A person approves (Where a person approves)Where offeredYours
6 of 10 vendors describe an approval step
Vendors describe approval before mitigation reaches live assets: a publish step, an approval gate, or sign-off thresholds.
Vicarius
“No script generated by ScriptAI can reach a live asset without a human publishing it first.”
Onit Security
“Every agent action passes an approval gate, writes to an audit trail, and is reversible in one step, across both remediations and mitigations.”
- Cogent Security
Zafran
“Keep humans in the loop with full command auditability and approval workflows at every step”
Qualys
“Condition-based patch blocking and the ServiceNow change integration keep every change under existing approval processes.”
Approval rules
If you buy: Where the product offers them. 6 of 10 vendors listed for this job describe an approval step on the pages read 19 Nov 2026.
If you build: You write which actions wait for a person.
Evidence (What evidence it leaves)Product recordsYours
6 of 10 vendors describe an audit trail
Vendors describe execution or audit logs of the match and the conclusions reached.
- Astelia
Onit Security
“Every conclusion traces back to the agent that produced it and the evidence it used.”
- Cogent Security
Zafran
“full command auditability”
Seemplicity
“an auditable trail of how the verdict was reached”
Logs
If you buy: What the product records. 6 of 10 vendors listed for this job describe an audit trail on the pages read 19 Nov 2026.
If you build: You log every step where the agent cannot edit it.
Model choice
If you buy: The supplier picks and updates the model.
Source
NCSC, Guidelines for secure AI system development: “You support users to evaluate and respond to model changes (for example by providing preview access and versioned APIs).”
If you build: You pick the model and can swap it.
Source
Anthropic, Scaling Managed Agents: “This allows the implementation of each to be swapped without disturbing the others.”
Each row opens to the vendor’s own words from its own page. All read 19 Nov 2026; counts from the vendors directory.
All 10 vendors for this job, lane by lane
A to Z, from each vendor’s own pages read 19 Nov 2026. Dot: stated; ring: none found.
- AsteliaAgentThe Hunter (Zero-Day Detection)Build or changeCustomers can build or change itApproval stepApproval step describedAudit trailAudit trail described
- Cogent SecurityAgentCogent AI agents (new disclosure)Build or changeNone foundApproval stepApproval step describedAudit trailAudit trail described
- Endor LabsAgentMalware Responder (AURI Agent)Build or changeCustomers can build or change itApproval stepNone foundAudit trailNone found
- MazeAgentMaze Cloud agentsBuild or changeNone foundApproval stepNone foundAudit trailNone found
- Onit SecurityAgentZero-Day AgentBuild or changeNone foundApproval stepApproval step describedAudit trailAudit trail described
- QualysAgentAgent Insta; Agent NyraBuild or changeNone foundApproval stepApproval step describedAudit trailNone found
- Rival SecurityAgent0-day responseBuild or changeCustomers can build or change itApproval stepNone foundAudit trailAudit trail described
- SeemplicityAgentSCA AnalystBuild or changeNone foundApproval stepNone foundAudit trailAudit trail described
- VicariusAgentScriptAIBuild or changeNone foundApproval stepApproval step describedAudit trailNone found
- ZafranAgentZero Day AgentBuild or changeCustomers can build or change itApproval stepApproval step describedAudit trailAudit trail described
None found for this job: approval step, 4 of 10 vendors (Endor Labs, Maze, Rival Security, Seemplicity); audit trail, 4 of 10 (Endor Labs, Maze, Qualys, Vicarius); customer can build or change the agent, 6 of 10.
Code and dependency fixes
An agent proposes a code change or a dependency upgrade, usually as a pull request.
Reads (What an agent reads)
4 of 16 vendors quoted on what the agent reads
Published agents read the finding, the repository source, dependency manifests and build or test results.
- Endor LabsSCA Remediation (AURI Agent)
- GitHubAgentic autofix for code scanning
BacklineCodie (Coder)
“Prepares clean, production-ready fixes for code and configurations.”
SemgrepSemgrep Agentic Workflows (fix step)
“Drive code changes to resolve issues. Generate PRs that reference impact, context, and validation to accelerate developer action.”
Does alone (What it does without a person)Supplier’s settingsYours
8 of 16 vendors let customers build or change the agent
Vendors describe agents that explore the code, generate a fix, validate it (build, tests) and open a pull request or merge request.
Endor LabsSCA Remediation (AURI Agent)
“It picks the version that fixes without breaking, makes the change, runs your build, and opens the PR.”
GitHubAgentic autofix for code scanning
“It explores the codebase, generates a fix, and validates the changes.”
SnykEvo Remediation Agent (dependency fixes)
“runs the full fix loop autonomously and at scale, across both newly written code and the backlog, starting with Snyk Open Source issues at launch.”
Rival SecurityAuto-remediate
“Plan, test and open the fix as a PR.”
Tools and permissions
If you buy: Inside the supplier’s settings. 8 of 16 vendors listed for this job state on their pages that customers can build or change the agent; for the other 8, none found on 19 Nov 2026.
Source
NIST AI RMF 1.0: “MAP 4: Risks and benefits are mapped for all components of the AI system including third-party software and data.”
If you build: You grant each tool and each credential.
A person approves (Where a person approves)Where offeredYours
12 of 16 vendors describe an approval step
Vendors describe a human review and merge step on the pull request, and in some products an approval before any file edit or branch push.
GitHub
“Always review and test the generated changes before you merge the pull request.”
Semgrep
“You remain in full control over reviewing and merging the PR or MR.”
Google
“Currently, all patches generated by CodeMender are reviewed by human researchers”
Endor Labs
“Require approval before any file edit, branch push, or pull request, and check every change against your policy: allow, block, or ask a human.”
Snyk
“A fix that might break the build is queued for approval, not merged, so Remediation Agent is built to deliver mergeable fixes.”
Checkmarx
“Developers review and merge each fix.”
Rival Security
“Rival plans the fix, tests it, and opens the pull request, so your team reviews instead of digging.”
Seemplicity
“Each option carries a Readiness Indicator showing whether it can be deployed immediately or needs review first.”
Approval rules
If you buy: Where the product offers them. 12 of 16 vendors listed for this job describe an approval step on the pages read 19 Nov 2026.
If you build: You write which actions wait for a person.
Evidence (What evidence it leaves)Product recordsYours
7 of 16 vendors describe an audit trail
Vendors describe the pull request itself as the record, plus agent logs and audit trails in some products.
GitHub
“If the agent can't generate a fix, it remains as an assignee, and you can click View Session on the alert timeline to review the agent's log.”
Backline
“Since every fix is delivered as a standard Pull Request, you can revert any change using your existing Git workflows and CI/CD rollback procedures”
Snyk
“every agent decision is captured in an audit trail.”
Checkmarx
“AppSec retains policy control and full traceability over every automated decision.”
Pixee
“Every triage decision is persisted with a timestamp, the classification outcome (true positive, false positive, won't fix), and the LLM justification that produced the verdict.”
Logs
If you buy: What the product records. 7 of 16 vendors listed for this job describe an audit trail on the pages read 19 Nov 2026.
If you build: You log every step where the agent cannot edit it.
Model choice
If you buy: The supplier picks and updates the model.
Source
NCSC, Guidelines for secure AI system development: “You support users to evaluate and respond to model changes (for example by providing preview access and versioned APIs).”
If you build: You pick the model and can swap it.
Source
Anthropic, Scaling Managed Agents: “This allows the implementation of each to be swapped without disturbing the others.”
Each row opens to the vendor’s own words from its own page. All read 19 Nov 2026; counts from the vendors directory.
All 16 vendors for this job, lane by lane
A to Z, from each vendor’s own pages read 19 Nov 2026. Dot: stated; ring: none found.
- AikidoAgentAikido Agent and AutoFixBuild or changeCustomers can build or change itApproval stepApproval step describedAudit trailNone found
- AverlonAgentRemediation AgentsBuild or changeNone foundApproval stepNone foundAudit trailNone found
- BacklineAgentCodie (Coder)Build or changeNone foundApproval stepApproval step describedAudit trailNone found
- CheckmarxAgentRemediation Assist; Developer AssistBuild or changeNone foundApproval stepApproval step describedAudit trailAudit trail described
- CorridorAgentCorridor Agent (fixes)Build or changeCustomers can build or change itApproval stepNone foundAudit trailNone found
- CycodeAgentFix & Remediation AgentBuild or changeCustomers can build or change itApproval stepApproval step describedAudit trailAudit trail described
- Endor LabsAgentSCA Remediation (AURI Agent)Build or changeCustomers can build or change itApproval stepApproval step describedAudit trailNone found
- GitHubAgentAgentic autofix for code scanning; Copilot Autofix; Copilot assigned to Dependabot alertsBuild or changeCustomers can build or change itApproval stepApproval step describedAudit trailAudit trail described
- GoogleAgentCodeMenderBuild or changeNone foundApproval stepApproval step describedAudit trailNone found
- MazeAgentMaze Code agentsBuild or changeNone foundApproval stepApproval step describedAudit trailNone found
- MobbAgentMobb AI AgentBuild or changeNone foundApproval stepNone foundAudit trailNone found
- PixeeAgentPixee fix generationBuild or changeCustomers can build or change itApproval stepApproval step describedAudit trailAudit trail described
- Rival SecurityAgentAuto-remediate; Dependency auditBuild or changeCustomers can build or change itApproval stepApproval step describedAudit trailAudit trail described
- SeemplicityAgentSAST AnalystBuild or changeNone foundApproval stepNone foundAudit trailAudit trail described
- SemgrepAgentSemgrep Agentic Workflows (fix step)Build or changeCustomers can build or change itApproval stepApproval step describedAudit trailNone found
- SnykAgentEvo Remediation Agent (dependency fixes)Build or changeNone foundApproval stepApproval step describedAudit trailAudit trail described
None found for this job: approval step, 4 of 16 vendors (Averlon, Corridor, Mobb, Seemplicity); audit trail, 9 of 16 (Aikido, Averlon, Backline, Corridor, Endor Labs, Google, Maze, Mobb, Semgrep); customer can build or change the agent, 8 of 16.
Cloud posture
An agent investigates cloud misconfiguration findings and proposes or applies a fix.
Reads (What an agent reads)
4 of 7 vendors quoted on what the agent reads
Published agents read cloud configuration findings, resource metadata, and in some products infrastructure-as-code changes.
TamnoonTami
“Tami, Tamnoon’s AI Cloud Security Agent, enriches each alert with context it captures from analyzing metadata and performing active investigation of your cloud environment.”
AverlonInfrastructure Risk PreCog Agent
“The Infrastructure Risk PreCog Agent evaluates proposed infrastructure changes before they are merged.”
UpwindGreen
“Implementation-ready changes across code, IaC, packages, images, and cloud configuration”
- WizGreen Agent (cloud misconfiguration remediation)
Does alone (What it does without a person)Supplier’s settingsYours
4 of 7 vendors let customers build or change the agent
Vendors describe agents that enrich alerts, rank misconfigurations and generate fixes; Wiz describes automatic remediation for “high-confidence verdicts”, and Prowler's assistant is read-only guidance.
WizGreen Agent (cloud misconfiguration remediation)
“Automatically trigger remediation actions when the Green Agent reaches a high-confidence remediation verdict, such as blocking public S3 access or patching a vulnerable workload”
SysdigPosture agent (Sysdig Agentic AI)
“Fixes misconfigurations”
ProwlerProwler Lighthouse AI
“Get step-by-step remediation for each finding: a plain explanation, the commands to run, and alternatives.”
- TamnoonTami
Tools and permissions
If you buy: Inside the supplier’s settings. 4 of 7 vendors listed for this job state on their pages that customers can build or change the agent; for the other 3, none found on 19 Nov 2026.
Source
NIST AI RMF 1.0: “MAP 4: Risks and benefits are mapped for all components of the AI system including third-party software and data.”
If you build: You grant each tool and each credential.
A person approves (Where a person approves)Where offeredYours
4 of 7 vendors describe an approval step
Vendors describe routing lower-confidence or high-impact remediation to a person, or running fixes through the customer's change process.
- Wiz
Sysdig
“high-impact actions come to your team for approval first”
Tamnoon
“Tamnoon proves each remediation safe and runs it through your own change process.”
Approval rules
If you buy: Where the product offers them. 4 of 7 vendors listed for this job describe an approval step on the pages read 19 Nov 2026.
If you build: You write which actions wait for a person.
Evidence (What evidence it leaves)Product recordsYours
4 of 7 vendors describe an audit trail
Vendors describe an investigation view, recorded reasoning for each action, or a retained decision record.
- Wiz
Sysdig
“Every action is recorded along with the reasoning behind it”
Tamnoon
“decision record retained, auditor-readable, attributable”
Logs
If you buy: What the product records. 4 of 7 vendors listed for this job describe an audit trail on the pages read 19 Nov 2026.
If you build: You log every step where the agent cannot edit it.
Model choice
If you buy: The supplier picks and updates the model.
Source
NCSC, Guidelines for secure AI system development: “You support users to evaluate and respond to model changes (for example by providing preview access and versioned APIs).”
If you build: You pick the model and can swap it.
Source
Anthropic, Scaling Managed Agents: “This allows the implementation of each to be swapped without disturbing the others.”
Each row opens to the vendor’s own words from its own page. All read 19 Nov 2026; counts from the vendors directory.
All 7 vendors for this job, lane by lane
A to Z, from each vendor’s own pages read 19 Nov 2026. Dot: stated; ring: none found.
- AverlonAgentInfrastructure Risk PreCog AgentBuild or changeNone foundApproval stepApproval step describedAudit trailNone found
- ProwlerAgentProwler Lighthouse AIBuild or changeCustomers can build or change itApproval stepNone foundAudit trailNone found
- Rival SecurityAgentCloud postureBuild or changeCustomers can build or change itApproval stepNone foundAudit trailAudit trail described
- SysdigAgentPosture agent (Sysdig Agentic AI)Build or changeCustomers can build or change itApproval stepApproval step describedAudit trailAudit trail described
- TamnoonAgentTamiBuild or changeNone foundApproval stepApproval step describedAudit trailAudit trail described
- UpwindAgentGreenBuild or changeNone foundApproval stepNone foundAudit trailNone found
- WizAgentGreen Agent (cloud misconfiguration remediation)Build or changeCustomers can build or change itApproval stepApproval step describedAudit trailAudit trail described
None found for this job: approval step, 3 of 7 vendors (Prowler, Rival Security, Upwind); audit trail, 3 of 7 (Averlon, Prowler, Upwind); customer can build or change the agent, 3 of 7.
Threat modeling
An agent builds or updates a threat model from designs, tickets, diagrams or code.
Reads (What an agent reads)
4 of 6 vendors quoted on what the agent reads
Published agents read design documents, diagrams, text prompts or the repository.
IriusRiskJeff (AI threat modeling assistant)
“Jeff AI is able to take text prompts or images from you to create a full threat model”
SeezoSeezo Agent and Seezo AI Plugin
“Seezo understands any design-level input and delivers a complete security design review and threat model in minutes”
- ThreatModelerNexusAI
SnykEvo Agentic Security Review
“An agent-driven security review of your repository: it builds an application and threat model, tests whether controls hold against credible threats, and delivers evidence-backed findings with clear remediation guidance.”
Does alone (What it does without a person)Supplier’s settingsYours
2 of 6 vendors let customers build or change the agent
Vendors describe agents that draft a threat model and a design security review without a person writing it first.
- SeezoSeezo Agent and Seezo AI Plugin
- IriusRiskJeff (AI threat modeling assistant)
ThreatModelerNexusAI
“Let NexusAI handle the groundwork of creating and analyzing threat models, identifying risks, and proposing updates”
Tools and permissions
If you buy: Inside the supplier’s settings. 2 of 6 vendors listed for this job state on their pages that customers can build or change the agent; for the other 4, none found on 19 Nov 2026.
Source
NIST AI RMF 1.0: “MAP 4: Risks and benefits are mapped for all components of the AI system including third-party software and data.”
If you build: You grant each tool and each credential.
A person approves (Where a person approves)Where offeredYours
3 of 6 vendors describe an approval step
Vendors describe the threat model as something the user reviews and approves before changes apply. Seezo describes no approval step; none found on 19 Nov 2026.
IriusRisk
“You stay in control of the threat model”
ThreatModeler
“Your team reviews and approves changes before they're applied, with an auditable record of decisions.”
Approval rules
If you buy: Where the product offers them. 3 of 6 vendors listed for this job describe an approval step on the pages read 19 Nov 2026.
If you build: You write which actions wait for a person.
Evidence (What evidence it leaves)Product recordsYours
5 of 6 vendors describe an audit trail
Vendors describe audit logs, traceability or a record of decisions for the model.
IriusRisk
“There is full traceability, audit logs, references to standards etc”
ThreatModeler
“with an auditable record of decisions”
Seezo
“Security teams get full visibility into what skills fired and when.”
Logs
If you buy: What the product records. 5 of 6 vendors listed for this job describe an audit trail on the pages read 19 Nov 2026.
If you build: You log every step where the agent cannot edit it.
Model choice
If you buy: The supplier picks and updates the model.
Source
NCSC, Guidelines for secure AI system development: “You support users to evaluate and respond to model changes (for example by providing preview access and versioned APIs).”
If you build: You pick the model and can swap it.
Source
Anthropic, Scaling Managed Agents: “This allows the implementation of each to be swapped without disturbing the others.”
Each row opens to the vendor’s own words from its own page. All read 19 Nov 2026; counts from the vendors directory.
All 6 vendors for this job, lane by lane
A to Z, from each vendor’s own pages read 19 Nov 2026. Dot: stated; ring: none found.
- IriusRiskAgentJeff (AI threat modeling assistant); MCP-connected AI agentsBuild or changeNone foundApproval stepApproval step describedAudit trailAudit trail described
- PixeeAgentForesight (design-stage review)Build or changeNone foundApproval stepApproval step describedAudit trailNone found
- Rival SecurityAgentThreat modelBuild or changeCustomers can build or change itApproval stepNone foundAudit trailAudit trail described
- SeezoAgentSeezo Agent and Seezo AI PluginBuild or changeCustomers can build or change itApproval stepNone foundAudit trailAudit trail described
- SnykAgentEvo Agentic Security ReviewBuild or changeNone foundApproval stepNone foundAudit trailAudit trail described
- ThreatModelerAgentNexusAI; System Mapping Agent, Graph Agent, Reporting AgentBuild or changeNone foundApproval stepApproval step describedAudit trailAudit trail described
None found for this job: approval step, 3 of 6 vendors (Rival Security, Seezo, Snyk); audit trail, 1 of 6 (Pixee); customer can build or change the agent, 4 of 6.
Access reviews
An agent prepares or performs access certification: which entitlements are still needed.
Reads (What an agent reads)
2 of 2 vendors quoted on what the agent reads
Published agents read entitlement and identity data and the usage and risk context around each review item.
- LumosAccess Review Agent (with Albus)
VezaAccess Review Agent (part of Veza Access Agents)
“Surface high-risk identities, and get explanations of complex reviews in plain language.”
Does alone (What it does without a person)Supplier’s settingsYours
1 of 2 vendors let customers build or change the agent
Lumos describes an agent that certifies low-risk access itself and surfaces only the items needing a human call. Veza describes an agent that highlights risk and explains review items, with reviewers still deciding.
LumosAccess Review Agent (with Albus)
“It certifies the safe access on its own and brings you only the items that need a human call.”
- VezaAccess Review Agent (part of Veza Access Agents)
Tools and permissions
If you buy: Inside the supplier’s settings. 1 of 2 vendors listed for this job state on their pages that customers can build or change the agent; for the other 1, none found on 19 Nov 2026.
Source
NIST AI RMF 1.0: “MAP 4: Risks and benefits are mapped for all components of the AI system including third-party software and data.”
If you build: You grant each tool and each credential.
A person approves (Where a person approves)Where offeredYours
2 of 2 vendors describe an approval step
Lumos describes customer-set policies and approval thresholds; Veza describes dual reviewer sign-off before certification completes.
Lumos
“You set the policies, guardrails, and approval thresholds.”
Veza
“Support for multiple reviewers who both must sign-off on an access review prior to completion and certification.”
Approval rules
If you buy: Where the product offers them. 2 of 2 vendors listed for this job describe an approval step on the pages read 19 Nov 2026.
If you build: You write which actions wait for a person.
Evidence (What evidence it leaves)Product recordsYours
1 of 2 vendors describe an audit trail
Lumos describes decisions stored as audit evidence. Veza's page: none found on 19 Nov 2026 for an agent audit trail.
Lumos
“every decision is stored as audit evidence”
Logs
If you buy: What the product records. 1 of 2 vendors listed for this job describe an audit trail on the pages read 19 Nov 2026.
If you build: You log every step where the agent cannot edit it.
Model choice
If you buy: The supplier picks and updates the model.
Source
NCSC, Guidelines for secure AI system development: “You support users to evaluate and respond to model changes (for example by providing preview access and versioned APIs).”
If you build: You pick the model and can swap it.
Source
Anthropic, Scaling Managed Agents: “This allows the implementation of each to be swapped without disturbing the others.”
Each row opens to the vendor’s own words from its own page. All read 19 Nov 2026; counts from the vendors directory.
All 2 vendors for this job, lane by lane
A to Z, from each vendor’s own pages read 19 Nov 2026. Dot: stated; ring: none found.
- LumosAgentAccess Review Agent (with Albus)Build or changeCustomers can build or change itApproval stepApproval step describedAudit trailAudit trail described
- VezaAgentAccess Review Agent (part of Veza Access Agents)Build or changeNone foundApproval stepApproval step describedAudit trailNone found
None found for this job: approval step, 0 of 2 vendors; audit trail, 1 of 2 (Veza); customer can build or change the agent, 1 of 2.
Every listed agent, four answers each
85 agents under 39 vendors. A vendor's approval and evidence text is printed once, on the vendor's row. Where a vendor published nothing, the cell says none found. Type a vendor, agent or job to filter.
| Vendor | Agents | Reads | Acts alone | Where a person approves | Evidence it leaves | Source |
|---|---|---|---|---|---|---|
| Aikido | Aikido Agent (CVE Exploitability Analysis) Aikido Agent and AutoFix | How the vulnerable package is used in your code, in a repository or container. A finding in chat, including its risk and attack paths. | Reads how a vulnerable package is used in a repository or container and ignores, downgrades, upgrades or snoozes the CVE finding with a reason. The agent can adapt fixes on a finding; AutoFix turns scanner findings into pull requests. | Each allowed action can be set to Fully Automated or Human Approval; with approval the agent waits for a person to approve it from the analysis history. | Each ignore, downgrade or upgrade carries an agent-provided reason, and runs can be reviewed in an analysis history. | Source, Source, read 19 Nov 2026 |
| Amplify Security | Amplify Console triage and remediation agents | Alerts from any source, weighed against the company priorities you set. | Takes alerts from any source, triages them against company priorities, and offers one-click remediations that can be customized. | None found on 19 Nov 2026 on the home, products or docs landing pages for an approval step. | None found on 19 Nov 2026 for a log of agent decisions; the products page only mentions audit-ready narrative reports. | Source, read 19 Nov 2026 |
| Astelia | The Investigator (Automated Triage) The Fixer (Agentic Remediation) The Hunter (Zero-Day Detection) | Each new vulnerability, with its reachability and operational impact. Each exposure, to choose a patch, firewall rule or IPS rule. Newly disclosed and zero-day vulnerabilities, checked against your SBOM and software. | Triages each new vulnerability by reachability and operational impact, validates it and coordinates next steps. Picks the fastest fix (patch, firewall or IPS rule) and carries it out through the existing stack with a person in the loop. Analyzes newly disclosed and zero-day vulnerabilities against the customer's SBOM and software to find what is reachable. | In the post's example workflow, a drafted ticket and a proposed firewall rule wait in an approval queue for a person to click. | Post says each step is recorded in an execution log and each decision in an audit log. | Source, read 19 Nov 2026 |
| Averlon | Mitigation Agents Remediation Agents Infrastructure Risk PreCog Agent | None found on 19 Nov 2026 None found on 19 Nov 2026 Proposed infrastructure changes, before they are merged. | Listed on the platform page as a component alongside remediation agents; the vendor does not further description of what they do. Generate fixes delivered into developer workflows; the home page shows a Remediation Agent proposing a Dockerfile change in a pull request. Evaluates proposed infrastructure changes before merge and flags new risk such as internet exposure of vulnerable hosts. | For vulnerability triage and remediation, code and dependency fixes: None found on 19 Nov 2026 for the other jobs this vendor is listed under. For cloud posture: For the Infrastructure Risk PreCog Agent, the finding sits on the pull request and the team can block or fix the change before production. For the Remediation Agents the pages read give delivery wording (ready-to-merge pull request) and a customer review-time figure, which name no approval step. | None found on 19 Nov 2026 on the platform page, home page or remediation blog post for an agent action log. | Source, Source, Source, read 19 Nov 2026 |
| Backline | Andy (Analyst) Polly (Planner) Vera (Verifier) Codie (Coder) | Vulnerabilities and the context of your environment. Risk, SLA and environment context for each vulnerability. Each proposed fix and its test results. The code and configurations that need a fix. | Interprets vulnerabilities, assesses risk and collects environment context. Plans remediation by risk, SLA and environment context. Tests each fix for safety and regressions. Prepares fixes for code and configurations. | For vulnerability triage and remediation: None found on 19 Nov 2026 for the other jobs this vendor is listed under. For code and dependency fixes: Product page says proposed fixes carry reasoning and validation for teams to review and approve, and engineers control what gets merged. | None found on 19 Nov 2026 for a log or trail of the agents' actions. The product FAQ says fixes are standard pull requests, so Git history and rollback apply; that is not counted. | Source, read 19 Nov 2026 |
| Checkmarx | Triage Assist Remediation Assist Developer Assist | Findings, with the reachability of each vulnerability. The vulnerabilities Checkmarx has already flagged as mattering. The code you write in an AI coding tool, plus context pulled from Checkmarx One. | Separates exploitable risk from severity noise using reachability. Generates merge-ready pull requests for the vulnerabilities that matter. Runs a find and fix loop in AI coding tools: detects, gets context, generates and verifies a fix before commit. | For vulnerability triage and remediation: None found on 19 Nov 2026 for the other jobs this vendor is listed under. For code and dependency fixes: Developers review and merge each fix. | Vendor says AppSec keeps full traceability over automated decisions; the vendor does not show the log itself. | Source, read 19 Nov 2026 |
| Cogent Security | Cogent AI agents (vulnerability lifecycle) Cogent AI agents (new disclosure) | Vulnerability data at each step: normalization, enrichment, risk assessment and remediation planning. Your software inventory, matched against a new disclosure. | Unnamed agents handle normalization, enrichment, risk assessment and remediation planning, then run fixes inside customer policy. Home page says the agents identify vulnerable software soon after a new disclosure, before scanners can. | Customers choose where agents act alone and where they pause for sign-off, by action type, severity or asset sensitivity; high-impact changes need human approval. | Pages say every agent action is logged with its reasoning and links to the recommendation behind it. | Source, Source, read 19 Nov 2026 |
| Corridor | Corridor Agent Corridor Agent (fixes) | Your existing vulnerability backlog. Findings already reviewed in your backlog. | Works through an existing vulnerability backlog, reviews findings and creates guardrails so the same issues do not return. Produces fixes in bulk for reviewed findings and opens them as pull requests. | None found on 19 Nov 2026 on the page read for an approval step; the sample text shows fixes opened as pull requests, which is not a stated approval setting. | The audit trail text on the page is about AI coding agents' activity; none found on 19 Nov 2026 tied to the backlog agent's decisions. | Source, read 19 Nov 2026 |
| CrowdStrike | Exposure Prioritization Agent (Charlotte AI, Falcon Exposure Management) | Exposure data, to judge what attackers are likely to exploit. | Ranks vulnerabilities by likelihood of exploitation using exploitability, adversary intelligence, attack path and asset context. | CrowdStrike says automated actions are configured by the team and can require human approval. | CrowdStrike lists execution traces and audit logs among governance controls for AgentWorks agents. | Source, read 19 Nov 2026 |
| Cycode | Exploitability Agent Fix & Remediation Agent | Whether a CVE is reachable in the application. Findings in your code, to write a pull-request-ready fix. | Checks whether a SAST or SCA violation is reachable and exploitable in the application and its runtime environment, so exploitable risk is separated from false positives. Generates pull-request-ready fixes with the reasoning behind them. | The workflows page says customers set which actions need human review and which run autonomously, with confidence thresholds on every agent and approval gates. | The workflows page states that review logs and audit trails are kept and that decision traces are auditable. | Source, read 19 Nov 2026 |
| Dux Security | Dux AI agents | Exploitability of vulnerabilities and asset context. | Unnamed agents do exploitability analysis, suggest lightweight mitigations and tag assets to speed owner-led remediation. | None found on 19 Nov 2026 on the home page. It says mitigations are ones customers deploy themselves. | None found on 19 Nov 2026 on the home page; the sitemap lists no other pages. | Source, read 19 Nov 2026 |
| Endor Labs | AI SAST Remediation (AURI Agent) Malware Responder (AURI Agent) SCA Remediation (AURI Agent) | A case file for each AI SAST finding, read against your source at the pinned commit. Live supply-chain malware intelligence and your Endor inventory. Your dependency versions and build, to pick a version that fixes the issue. | Reads a case file for each AI SAST finding against the source, decides what is real, then writes the fix or files an exception. Matches supply-chain malware intelligence to the customer's inventory to say whether they are exposed, with affected projects and containment guidance. Chooses a dependency version that fixes the issue without breaking call paths, makes the change, runs the build and opens the pull request. | For vulnerability triage and remediation, code and dependency fixes: The remediation page says approval can be required before any file edit, branch push or pull request; the workflows page says agents are read-only by default with approval gates. For response to a new disclosure: None found on 19 Nov 2026 for an approval step on Malware Responder: the workflows page says most agents are read-only and cannot edit files, open pull requests or dismiss findings, and that only SCA Remediation and AI SAST Remediation change state, each behind a separate approval gate. | The Agents Hub usage view is stated to be metering, not a compliance audit log. A docs summary line says agent actions are attributed to a person; the page read does not show that record. | Source, read 19 Nov 2026 |
| GitHub | Assign security campaign alerts to Copilot Agentic autofix for code scanning Copilot Autofix Copilot assigned to Dependabot alerts | None found on 19 Nov 2026 The codebase around an assigned code scanning alert. One supported code scanning alert and the code it points to. The Dependabot alert assigned to it. | From a security campaign, 1 to 25 code scanning alerts can be assigned to Copilot in one session producing one pull request. Copilot explores the codebase for assigned code scanning alerts, writes and validates a fix, and opens a draft pull request. Generates one suggested code change for one code scanning alert that a developer reviews and applies. A Dependabot alert can be assigned to Copilot (or enabled third-party agents), which opens a draft pull request with a proposed fix. | Agentic autofix output is a draft pull request, and the user who assigns the alerts is added as a requested reviewer. GitHub tells users to review and test before merging. | When an alert is assigned to an agent, a session is created, and the agent's log can be reviewed from the alert timeline. | Source, Source, Source, read 19 Nov 2026 |
CodeMender | Code vulnerabilities and their root causes in source code. | A Google DeepMind research agent that finds root causes of code vulnerabilities, writes patches, validates them, and sends only vetted patches to human reviewers. | At the time of the post, a human researcher reviewed every CodeMender patch before it went upstream; automated critique agents review first. | None found on 19 Nov 2026 for an audit log product feature. The post shows example agent reasoning steps, but as research illustrations. | Source, read 19 Nov 2026 | |
| IriusRisk | Jeff (AI threat modeling assistant) MCP-connected AI agents | Text prompts or images that describe the system. None found on 19 Nov 2026 | Turns text prompts or images into a threat model diagram inside IriusRisk, which the user then reviews and edits. External AI agents can be connected to IriusRisk's threat modeling process for modeling and analysis. | The user reviews and can edit the generated diagram before it proceeds; the human is responsible for judging response quality. | Threats and countermeasures come from the rules engine rather than the AI, and the page says this carries traceability and audit logs. | Source, Source, read 19 Nov 2026 |
| Lumos | Access Review Agent (with Albus) | Access in a review campaign, to judge which items are safe. | Scopes and runs access review campaigns, certifies access it judges safe, and sends the rest to a human with evidence. | Reviewers approve, deny or delegate flagged items; policies and approval thresholds are set by the customer. | Decisions are stored as audit evidence, with reports formatted for SOC 2, SOX and ISO 27001. | Source, read 19 Nov 2026 |
| Maze | Maze remediation agents Maze Cloud agents Maze Code agents | Vulnerabilities, to find the root cause. CVEs in containers and VMs. Where a vulnerability came from in the codebase, and who owns it. | Find root cause, validate a fix and route it to the responsible developer or coding agent. Triage and remediate container and VM CVEs and, per the page, catch zero-days before scanners see them. Write code fixes that match the codebase and ship the PR to the owner. | For vulnerability triage and remediation, code and dependency fixes: Remediation page says developers review and merge fixes like any PR, and the team controls what gets merged. For response to a new disclosure: None found on 19 Nov 2026 for the other jobs this vendor is listed under. | None found on 19 Nov 2026 for an agent action log. Pages say each fix and recommendation carries its evidence and reasoning. | Source, Source, Source, read 19 Nov 2026 |
| Microsoft | Vulnerability Remediation Agent (Security Copilot in Intune, public preview) | Microsoft Defender Vulnerability Management data for managed Windows devices and apps. | Reads Microsoft Defender Vulnerability Management data, ranks CVEs on managed Windows devices and apps, and gives step-by-step Intune guidance that an admin carries out. | Intune says its agents act with administrator oversight and review. For this agent the fix itself is done by a person in Intune from the agent's suggestion; no agent-side approval gate for the vulnerability agent was found. The separate Device Offboarding Agent does require admin approval. | None found on 19 Nov 2026 for an audit log of the vulnerability agent's actions. A person can mark a suggestion applied, which gives the agent a record of past fixes across runs; Microsoft Entra agent pages separately describe activity history for Entra agents. | Source, read 19 Nov 2026 |
| Mobb | Mobb AI Agent | Code vulnerabilities reported by scanners. | Generates ready-to-merge pull requests with fixes for scanner-reported code vulnerabilities. | None found on 19 Nov 2026 for an approval step; the page says developers apply and commit the fixes, which is not a stated approval setting. | None found on 19 Nov 2026. | Source, read 19 Nov 2026 |
| Onit Security | Exploitability Agent Ownership Agent Remediation Agent Mitigation Agent Zero-Day Agent | None found on 19 Nov 2026 None found on 19 Nov 2026 None found on 19 Nov 2026 None found on 19 Nov 2026 A newly disclosed critical vulnerability, run against your environment. | Checks host conditions to judge whether a finding can be exploited in the customer's environment. Resolves the real owner from inventories, ticketing history, directories and repositories. Builds a step-by-step remediation plan; the pages describe a plan, not an opened pull request. Finds interim mitigations such as disabling a feature or tightening a firewall rule. Picks up newly disclosed critical vulnerabilities and returns exposed assets, a per-asset exploitability conclusion and recommended actions. | Platform page says every agent action passes an approval gate; the Zero-Day post says a person approves the decision before disruptive action. | Same platform page sentence says actions write to an audit trail; the AI agents post says conclusions trace back to the producing agent and evidence. | Source, Source, read 19 Nov 2026 |
| Pixee | Pixee triage engine (AI agents tier) Pixee fix generation Foresight (design-stage review) | Ambiguous findings, by investigating the dynamic codebase. Findings in your code, to write fixes. The design intent, before the code is written. | Sends each finding through deterministic checks first, then AI agents that investigate the codebase for ambiguous cases, and records a justification. Creates fixes as pull requests, from deterministic codemods and constrained AI generation, checked before delivery. Reads specs, ties findings to a spec line and a threat, and checks shipped code against the design promises. The page presents it as a system, not by the word agent. | For vulnerability triage and remediation: None found on 19 Nov 2026 for this job. For code and dependency fixes, threat modeling: Fixes arrive only as pull requests that go through existing review; the FAQ says there is no mode that allows direct commits. For design-stage review, Foresight ties each finding to a spec line that the user confirms or dismisses. | Each triage decision is stored with a timestamp, outcome and the model's justification; each fix is a traceable pull request. | Source, Source, read 19 Nov 2026 |
| ProjectDiscovery | Vuln Triage Agent (Neo) Verification Agent (Neo) | Reports from HackerOne, Bugcrowd, GitHub Security Advisories, mailboxes, scanner output, manual input and imported findings. Findings already raised, which it re-tests independently. | Reviews and tests incoming reports, advisories and scanner output to judge exploitability and severity before anyone acts on them. Independently re-tests findings after a testing run and files only the confirmed ones. | None found on 19 Nov 2026 on the Vuln Triage Agent page or the agent architecture page for a human approval step before an agent acts. | The homepage says every agent action is logged and can be queried and exported to a SIEM. | Source, Source, read 19 Nov 2026 |
| Prowler | Prowler Lighthouse AI | Findings from Prowler scans. | Answers questions about posture findings, explains risk and gives step-by-step remediation guidance; read only. | Not applicable by design: the vendor says the agent has no tools to change the environment and the team applies changes. | None found on 19 Nov 2026 on the Lighthouse AI page or homepage. | Source, read 19 Nov 2026 |
| Qualys | Agent Val Agent Sara Agent Insta Agent Nyra | Exposures in your environment, tested with TruConfirm. Remediation status of a vulnerability, beyond the ticket state. New advisories, live inventory and threat context. Active threat actor campaigns and your impacted assets. | Decides what to validate next, safely tests whether an exposure is reachable and weaponizable, then triggers mitigations and re-checks. Runs remediation in waves gated by a patch reliability score, with patchless options, rollback and closure verification. Correlates new advisories with live inventory and existing telemetry to produce scanless findings soon after a disclosure. Tracks active threat campaigns, maps weaponized CVEs to the customer's impacted assets and re-ranks the backlog. | Patches likely to succeed deploy with zero-touch automation, less certain ones go through ring-based validation, and the ServiceNow change integration keeps every change under existing approval processes. | Agent Val is described as keeping an audit trail when it re-checks exploit validation after mitigation. | Source, read 19 Nov 2026 |
| Rival Security | Triage findings 0-day response Auto-remediate Dependency audit Cloud posture Threat model | Scanner findings. Your stack, matched against a new CVE. The finding and the code or dependency it affects. Packages and their versions. Cloud misconfigurations. A repository, to map its attack surface. | Built-in agent that cuts scanner output down to the findings judged exploitable. Built-in agent that finds where a newly disclosed CVE affects the customer's stack. Built-in agent that plans and tests a fix and opens it as a pull request. Built-in agent that checks packages for risky versions. Built-in agent that ranks cloud misconfigurations by risk. Built-in agent that maps the attack surface of a repository. | For vulnerability triage and remediation, response to a new disclosure, cloud posture, threat modeling: None found on 19 Nov 2026 for an approval step specific to this job's agent (for triage, for cloud posture and for the threat model agent, the pages describe none); the home page says every agent action is 'traced, auditable and held to your policies'. For code and dependency fixes: Rival says it opens the fix as a pull request so your team reviews it, which applies to the fix and new-disclosure agents only. | Home page states each agent action is traced and auditable and held to the customer's policies. | Source, read 19 Nov 2026 |
| Seemplicity | Host VM Analyst SCA Analyst SAST Analyst | Host vulnerability findings and the conditions required for them to be exploitable. SBOM data, for zero-day response. SAST findings, scoped to their blast radius. | Checks exploit prerequisites against live host settings and lowers priority when a condition is missing. Checks whether vulnerable library functions are actually called, supporting SBOM-driven response to zero-days. Confirms a code-level finding is reachable and supplies a PR-ready fix. | Each Response Option carries a Readiness Indicator showing whether it can be deployed immediately or needs review first. The pages read state no approval gate on the analysts' conclusions. | The article says each finding carries a record of how the conclusion was reached. | Source, Source, read 19 Nov 2026 |
| Seezo | Seezo Agent and Seezo AI Plugin | Design-level input, such as requirements and architecture. | Builds a design-level security review and threat model from design artifacts, and works inside coding tools through a plugin. | The threat modeling page lists a triage step but describes no approval or sign-off workflow. The plugin page says drift between requirements and what was built is reported. | Vendor says each threat cites its source and standard, and security teams see which skills fired and when. | Source, read 19 Nov 2026 |
| Semgrep | Semgrep Multimodal (formerly Semgrep Assistant) Semgrep Agentic Workflows (review step) Semgrep Agentic Workflows (fix step) | Your code, its languages and libraries, and your triage history. Potential vulnerabilities, in context. Issues already found by Semgrep, with their impact and context. | Auto-analyzes Semgrep Code findings and recommends which can safely be ignored, with an explanation for each recommendation. The docs call it AI-driven, not an agent. A workflow step that reviews potential vulnerabilities in context for validity and priority and captures feedback. A workflow step that drives code changes and generates pull requests that cite impact, context and validation. | For vulnerability triage and remediation: None found on 19 Nov 2026 for the other jobs this vendor is listed under. For code and dependency fixes: Autofix opens a draft pull request or merge request and leaves review and merge to the team. Multimodal ignore suggestions are recommendations. | None found on 19 Nov 2026 for a log of the agents' decisions. The vendor calls custom workflows auditable; that is not counted. | Source, Source, read 19 Nov 2026 |
| Snyk | Evo Remediation Agent Evo Remediation Agent (dependency fixes) Evo Agentic Security Review | New code and the existing backlog, in a sandbox. Snyk Open Source issues, new code and the backlog. Your repository, to build an application and threat model. | Runs a scan, fix and validate loop in a sandbox on new code and on the existing backlog, with people owning the merge decision. Produces fixes for open source dependency issues first; fixes that might break the build wait for approval rather than merging. Reviews a repository, builds an application model and a threat model from code and context, and reports evidence-backed findings. | For vulnerability triage and remediation, code and dependency fixes: Remediation is gated: a fix that might break the build is queued for approval instead of merging. Prevention inside a coding agent is stated to need no human sign-off. For threat modeling: None found on 19 Nov 2026 for an approval step specific to Evo Agentic Security Review; the queued-for-approval gate belongs to Evo Remediation Agent. | Vendor says every agent decision lands in an audit trail; the vendor does not show what the trail contains or how to export it. | Source, Source, read 19 Nov 2026 |
| Sysdig | Posture agent (Sysdig Agentic AI) | None found on 19 Nov 2026 | Fixes cloud misconfigurations and builds custom policies, with high-impact actions routed to the team for approval. | Routine analysis runs on its own; high-impact actions wait for team approval. | Each action is recorded with its reasoning. | Source, read 19 Nov 2026 |
| Tamnoon | Tami | Alert metadata and your cloud environment, by active investigation. | An AI cloud security agent that enriches each CNAPP alert by investigating the cloud environment, then rates the fix safe, risky or awaiting data. | Fixes run through the customer's own change process; unsafe ones are declined in writing. | Each decision is kept as a record the vendor calls auditor-readable and attributable. | Source, read 19 Nov 2026 |
| Tenable | Tenable Hexa AI | Cross-domain vulnerabilities. | An orchestrator over a fleet of agents that triages vulnerabilities on a schedule, creates tickets, connects to patch tools and triggers rescans when tickets close. | The data sheet says approvals by a human are built in; Patch Management also has policy controls for approvals. | None found on 19 Nov 2026 for agent actions on the Hexa AI page, data sheet or Patch Management page. | Source, read 19 Nov 2026 |
| ThreatModeler | NexusAI System Mapping Agent, Graph Agent, Reporting Agent | Threat models and their risks. None found on 19 Nov 2026 | Creates threat models from descriptions or existing materials, finds risks and proposes updates as architecture changes. Three further agents are named on the homepage; the homepage gives no description of what each does. | Proposed changes are applied only after the team reviews and approves them. | An auditable record of decisions is stated. | Source, read 19 Nov 2026 |
| Torq | Torq agentic AI in vulnerability management (HyperAgents) | Live threat intelligence, used to adjust prioritization models. | Torq's blog describes agentic AI that adjusts prioritization from live threat intelligence, reassigns work on delays, and coordinates patch, ticket and verification steps through connected tools. The blog does not name a specific vulnerability agent. | The builder deploys nothing until a person approves it. Per-action approval for vulnerability remediation was not described. | Torq says each reasoning step, conclusion and action of a HyperAgent is logged. | Source, read 19 Nov 2026 |
| Upwind | Green | Code, IaC, packages, images and cloud configuration. | Looks across all findings for the few changes that remove the most risk and produces fixes as pull requests across code, IaC and cloud configuration. | None found on 19 Nov 2026 on the agentic security page; fixes are delivered as pull requests, which a team reviews in its normal flow (inference, not stated). | None found on 19 Nov 2026. | Source, read 19 Nov 2026 |
| Veza | Access Review Agent (part of Veza Access Agents) | Identities and complex access reviews. | Gives reviewers AI-assisted recommendations, surfaces high-risk identities and explains complex reviews in plain language; it assists reviewers rather than certifying alone. | Multi-level reviewer sign-off is a listed feature; the page does not say whether automated actions always get human review. | The pages mention audit readiness and an API for certifications but describe no audit trail for agent actions. | Source, read 19 Nov 2026 |
| Vicarius | Agentic Exploit Simulation (vIntelligence) ScriptAI | Each vulnerability in your live environment. Vulnerabilities being actively exploited that have no vendor fix yet. | Tests each vulnerability in the live environment to confirm an attacker can exploit it before it reaches the team. Generates detection and remediation scripts automatically when an exploited vulnerability has no patch, or on request in plain language. | For vulnerability triage and remediation: None found on 19 Nov 2026 for an approval step on Agentic Exploit Simulation itself; the draft-then-publish step belongs to ScriptAI, which covers the remediation scripts. For response to a new disclosure: Generated scripts are drafts and cannot reach assets until a person publishes them. | None found on 19 Nov 2026 for ScriptAI actions; the navigation mentions audit-ready evidence for GRC teams only. | Source, Source, read 19 Nov 2026 |
| Wiz | Green Agent Green Agent (cloud misconfiguration remediation) | Security issues in your cloud, as an investigator. Cloud misconfigurations and vulnerable workloads, such as public S3 access. | Investigates high-risk Issues in Wiz, finds root cause and owner, gives a Remediate or Ignore conclusion with confidence, and produces fix steps, pull requests or hand-offs to coding agents. In Wiz Workflows, high-confidence Green Agent conclusions can trigger actions such as blocking public S3 access; lower-confidence ones go to a person. | Lower-confidence Green Agent conclusions can be routed to a developer in Slack for approval before remediation. | The investigation steps are laid out in an Investigation tab so teams can see how a recommendation was derived. | Source, Source, read 19 Nov 2026 |
| Zafran | Agentic Remediation Zero Day Agent | Enterprise vulnerabilities, with human oversight. A new disclosure, matched to your live software inventory. | Agents research CVE conditions, validate exploitability on live assets, and generate remediation scripts by simulating patch impact, with human oversight. Matches a pre-announced flaw to live software inventory to show which assets run the vulnerable version, are reachable, or are blocked by a control. | Vendor says approval workflows apply at every step of Agentic Remediation. | Same sentence claims full command auditability. | Source, Source, read 19 Nov 2026 |
Worked example: one agent against the four questions
Take Checkmarx's Triage Assist in vulnerability triage and remediation. First question, what it reads: reports of possible vulnerabilities arriving from bug bounty platforms, GitHub advisories, shared mailboxes and scanners, as the agent page lists them. Second, what it does without a person: it tests each report and forms its own view of how exploitable and how severe the issue is, before anyone acts. Third, where a person approves: the agent page and the architecture page describe no human approval step, so the answer on 19 Nov 2026 is none found. Fourth, what evidence it leaves: the vendor's homepage says each agent action is logged, can be queried and can be sent to a SIEM. One of the four answers is a gap, so ask Checkmarx where a person can stop this agent before it acts on your findings. Repeat the walk for any vendor on the map.
Sources for this example, each read 19 Nov 2026:
- Triage Assist, agent page, read 19 Nov 2026
- Checkmarx, evidence left, read 19 Nov 2026
Standards and guidance behind the four questions
Printed once for all six jobs. Read 19 Nov 2026.
- What it does without a person, OWASP LLM06:2025 Excessive Agency: “The root cause of Excessive Agency is typically one or more of: excessive functionality; excessive permissions; excessive autonomy.”
- What it does without a person, Anthropic, Building effective agents: “The autonomous nature of agents means higher costs, and the potential for compounding errors. We recommend extensive testing in sandboxed environments, along with the appropriate guardrails.”
- Where a person approves, OWASP LLM06:2025 Excessive Agency: “Utilise human-in-the-loop control to require a human to approve high-impact actions before they are taken.”
- Where a person approves, EU AI Act, Article 14: “or interrupt the system through a ‘stop’ button or a similar procedure that allows the system to come to a halt in a safe state.” Article 14 sets oversight duties for high-risk AI systems only; whether a given security agent is high-risk depends on its use.
- What evidence it leaves, MITRE ATLAS mitigations: “When deploying AI agents, implement logging of the intermediate steps of agentic actions and decisions, data access and tool use, installation commands, and identity of the agent.”
Read 19 Nov 2026. 85 named agents, counted as one per name a vendor lists for a job (names are separated by semicolons in the vendor's list; ThreatModeler's "System Mapping Agent, Graph Agent, Reporting Agent" is one listed name); 69 vendor-job listings; 39 vendors, each counted once. Quotes are each vendor’s own words about its own product.
Sources
- OWASP, LLM06:2025 Excessive Agency, read 19 Nov 2026
- MITRE ATLAS data, mitigation AML.M0030, read 19 Nov 2026
- EU AI Act, Article 14 (human oversight of high-risk AI systems), read 19 Nov 2026
- Anthropic, Building effective agents, read 19 Nov 2026
Last reviewed 19 Nov 2026