Agentic Index
Enclave vs Vulnetic (2026)
Both vendors last verified
Enclave and Vulnetic both run AI agents that pentest and report on the customer's systems, and they differ on breadth and on where they run. That verdict is the Agentic Index coverage score, graded from each vendor's own published materials.
Enclave, recently out of stealth, fields a team of agents that scan code, pentest live applications, review configurations, monitor CVEs, triage and prepare verified fixes, taking jobs in Slack or Teams and running on the customer's choice of frontier or open weight models. Vulnetic's Sable pentests web, API, network, cloud and Active Directory targets, follows the customer's own methodology, preserves every command and exploit trace, and can run on premises or air gapped. Neither has a public price. Choose Enclave for a broad agent team you task from chat. Choose Vulnetic for traceable pentests on premises.
On the Agentic Index AI SOC ranking, neither Enclave nor Vulnetic clears the bar, which asks for all five investigation loop capabilities documented in full. Enclave does not document observability and auditability in full, nor human oversight and guardrails; Vulnetic documents two of the five in full. 25 of the 86 vendors in the lane clear it. See the AI SOC ranking
This comparison is published by Agentic Index, an independent agentic AI vendor research platform. Enclave and Vulnetic are each graded against the same 14 capability Agentic Index taxonomy, from the vendor's own public materials under the Agentic Index verification standard, alongside 1,084 researched vendors. No vendor pays for placement and no vendor has reviewed this page. How this evidence is graded
Choose Enclave if
- One agent team should cover code scanning, pentests, configuration review and CVE monitoring.
- You want to hand out jobs in Slack or Teams.
- You want the agents on frontier or open weight models of your choosing.
Choose Vulnetic if
- Pentests must run on premises or air gapped.
- Every command and exploit trace should be kept.
- The agent should follow your own testing methodology.
| Feature | E Enclave |
V Vulnetic |
|---|---|---|
| Action & orchestration | ||
|
Integrations & Tool Calling Ability to connect agents to real systems through native integrations, OAuth-authenticated actions, custom tools, APIs, webhooks, or MCP-compatible tools. |
||
|
EnclaveIntegrations & Tool Calling Enclave takes jobs and reports evidence in Slack or Microsoft Teams, works from GitHub (pull request review and commit analysis), AWS Security Hub and scanners such as Wiz, Snyk, Semgrep, Dependabot and Socket, and prepares fixes as pull requests and tickets. In one example it flags a Terraform change in a pull request that would let every AWS account restore a production snapshot. SourceEnclave, enclave.airead 2026-10-06 |
||
|
VulneticIntegrations & Tool Calling Sable runs real tools and exploits against the customer's web applications and APIs, internal networks and Active Directory, using the credentials, URLs and API endpoints the team provides. Vulnetic mentions integrations for enterprise security teams without naming any, and it names no integration catalog or action framework. SourceVulnetic, vulnetic.ai, /web-application-pentesting and /enterpriseread 2026-10-06 |
||
|
Workflow Orchestration Ability to sequence, branch, retry, route, and combine deterministic workflow nodes with autonomous agent steps. |
||
|
EnclaveWorkflow Orchestration Enclave offers "an autonomous team of security agents that finishes scoped security jobs end to end". Specialist agents for code, infrastructure and live application testing run in parallel, Triage verifies and reproduces findings, and Remediation prepares fixes and retests them after deployment. The agents keep working when people sign off. SourceEnclave, enclave.airead 2026-10-06 |
||
|
VulneticWorkflow Orchestration Sable plans and executes authorized pentests following methodologies the customer writes, reusable instructions for reconnaissance, exploitation, validation, reporting and engagement specific testing. In Active Directory it finds and exploits paths to Domain Admin while the team controls scope and tradecraft. Engagements run in parallel, and on premises pricing is scoped by the number of parallel agents. SourceVulnetic, vulnetic.ai/methodologies, /active-directory-pentesting, /pentesting-platform-for-mssps and /on-premisesread 2026-10-06 |
||
|
Triggers & Channel Coverage How agents wake up and where they work: schedules, webhooks, message events, CRM events, inbox events, chat, email, voice, and collaboration tools. |
||
|
EnclaveTriggers & Channel Coverage Pull request review and commit analysis on GitHub, CVE Monitoring and continuous penetration testing wake the agents on repository events and new disclosures, besides jobs assigned in Slack or Teams. Enclave does not say how the events are wired. SourceEnclave, enclave.airead 2026-10-06 |
||
|
VulneticTriggers & Channel Coverage Teams launch a targeted retest on an existing vulnerability after remediation, a release or an infrastructure change, and run repeatable assessments as applications, APIs and internal environments change. Vulnetic names no schedule or event that starts Sable without a person. SourceVulnetic, vulnetic.ai/continuous-penetration-testingread 2026-10-06 |
||
| Knowledge & context | ||
|
Knowledge Grounding & RAG Ability to ground agent behavior in company data through document ingestion, retrieval, external knowledge APIs, semantic search, or RAG layers. |
||
|
EnclaveKnowledge Grounding & RAG The agents work from the customer's code, infrastructure, scanners and security tools, and a Network Inventory agent maps the attack surface. Enclave describes no maintained knowledge store the agents retrieve from. SourceEnclave, enclave.airead 2026-10-06 |
||
|
VulneticKnowledge Grounding & RAG Reconnaissance on each engagement grounds testing in the target environment, and the customer's methodologies guide Sable through it. Vulnetic names no maintained knowledge base or retrieval structure. SourceVulnetic, vulnetic.ai/methodologies and /ai-pentesting-agent; vulnetic.ai/ai-pentesting-agentread 2026-10-06 |
||
|
Memory & State Persistence Ability to persist context across a run, conversation, workflow, user, team, or longer-term memory layer. |
||
|
EnclaveMemory & State Persistence Each job starts from the scope a person sets in Slack or Teams. Enclave describes no memory its agents keep from one job to the next, what it would hold or how long it would last. SourceEnclave, enclave.airead 2026-10-06 |
||
|
VulneticMemory & State Persistence The reusable methodologies are playbooks the customer writes, and a retest keeps its new evidence with the original finding. Vulnetic does not say whether Sable keeps a memory of its own. SourceVulnetic, vulnetic.ai/methodologies and /continuous-penetration-testingread 2026-10-06 |
||
| Control & trust | ||
|
Human Oversight & Guardrails Approval steps, consent checkpoints, escalation rules, structured guardrails, policy constraints, and pause/resume controls. |
||
|
EnclaveHuman Oversight & Guardrails Jobs are scoped by the customer in Slack or Teams, evidence is reviewed there, and "a draft fix goes to review" as a pull request, so the team decides what merges through its own code host. Enclave names no approval step inside the product before an agent acts. SourceEnclave, enclave.airead 2026-10-06 |
||
|
VulneticHuman Oversight & Guardrails Teams define scope and rules of engagement, inspect and steer execution in natural language, and Sable keeps security teams in control from the first task through validated findings and remediation, with practitioner review in the same workflow. Steering and scope constrain the agent, but Vulnetic names no approval step before an exploit or other action goes through. SourceVulnetic, vulnetic.ai/ai-pentesting-agent, vulnetic.ai and /web-application-pentestingread 2026-10-06 |
||
|
Security, Identity & Governance RBAC, SSO, auditability, encryption, least-privilege tool access, compliance posture, and data handling policy. |
||
|
EnclaveSecurity, Identity & Governance Enclave displays an AICPA SOC 2 badge and lists a trust center at trust.enclave.ai. It names no SSO or role model. SourceEnclave, enclave.ai and trust.enclave.airead 2026-10-06 |
||
|
VulneticSecurity, Identity & Governance Vulnetic names no security attestation or access controls. SourceVulnetic, vulnetic.ai, /security and /enterpriseread 2026-10-06 |
||
|
Observability & Auditability Traces, logs, execution histories, metrics, audit events, and debugging detail for production agent behavior. |
||
|
EnclaveObservability & Auditability Enclave traces the evidence behind each finding and records verification results. After one fix was deployed, for example, it replayed the expired webhook request and recorded the HTTP 401 response. That evidence shows what the agents concluded. Enclave describes no run trace of the agents' own steps and tool calls. SourceEnclave, enclave.airead 2026-10-06 |
||
|
VulneticObservability & Auditability Sable will "preserve every decision and command" and its exploit traces so teams can "audit every action", and it records the evidence needed to reproduce each finding, beside validated findings, execution traces and executive ready reports. Retest evidence stays connected to the original finding. SourceVulnetic, vulnetic.ai/ai-pentesting-agent, /web-application-pentesting, /enterprise and /continuous-penetration-testingread 2026-10-06 |
||
|
Deployment & Data Residency Deployment modes and options, including SaaS, dedicated cloud, VPC, on-prem, hybrid, local runtime, and self-hosting. |
||
|
EnclaveDeployment & Data Residency Open weight models run on US inference providers, and customers can use Enclave's own inference or bring their own keys. Beyond where inference runs, Enclave publishes no hosting region for the platform itself and no way to run it in the customer's own environment. Its trust center is at trust.enclave.ai. SourceEnclave, enclave.ai; trust.enclave.airead 2026-10-06 |
||
|
VulneticDeployment & Data Residency On premises deployment supports private cloud, customer data center, hybrid and air gapped environments, with pricing scoped to the environment, testing volume and parallel agents. Sable can be placed inside an authorized network, a cloud VPC or a lab with access to internal targets, and it tests internal networks through a controlled relay. SourceVulnetic, vulnetic.ai/on-premises and /active-directory-pentestingread 2026-10-06 |
||
| Solution readiness | ||
|
Prebuilt Agents, Templates & Packs Ready-made workflows, packaged employees, templates, blueprints, industry solutions, and role-specific agents that reduce time-to-value. |
||
|
EnclavePrebuilt Agents, Templates & Packs Eight named specialist agents have stated jobs, covering Code Scanning, Pen Testing, Config Review, Network Inventory, CVE Monitoring, Intrusion Detection, Triage and Remediation. Forward deployed engineers also "build custom agents for your environment". SourceEnclave, enclave.airead 2026-10-06 |
||
|
VulneticPrebuilt Agents, Templates & Packs Customers create their own reusable methodology instructions for reconnaissance, exploitation, validation and reporting, and MSSPs standardize a methodology and deliver the work under their own brand. Vulnetic names no prebuilt methodology library and no agents beyond Sable. SourceVulnetic, vulnetic.ai/methodologies and /pentesting-platform-for-msspsread 2026-10-06 |
||
| Platform extensibility | ||
|
Model Flexibility & Routing Ability to work across multiple foundation models, route tasks to different models, or let buyers bring their own providers and keys. |
||
|
EnclaveModel Flexibility & Routing "Choose frontier models or open weights running on US inference providers. Use Enclave inference or bring your own keys." Models on offer include GPT-6 Astra, Luna and Sol and Claude Opus 5.5 and Claude Sonnet 5, OpenAI and Anthropic frontier models, beside open weight options, so customers choose among providers and can bring their own keys. SourceEnclave, enclave.airead 2026-10-06 |
||
|
VulneticModel Flexibility & Routing No model provider is named, even for the on premises and air gapped deployments, and Vulnetic names no model choice or routing for customers. SourceVulnetic, vulnetic.ai/ai-pentesting-agent and /on-premisesread 2026-10-06 |
||
|
APIs, SDKs & MCP Extensibility Composability layer: stable APIs, SDKs, MCP tool consumption/serving, custom tools, and integration into internal systems. |
||
|
EnclaveAPIs, SDKs & MCP Extensibility No API, SDK, MCP server or developer documentation is published. Custom agents are built by Enclave's forward deployed engineers, who embed with the customer's team and stay until the team runs the agents on its own. SourceEnclave, enclave.airead 2026-10-06 |
||
|
VulneticAPIs, SDKs & MCP Extensibility No API, SDK or MCP server for building on Vulnetic is published. The integrations Vulnetic mentions for enterprise security teams are not named. SourceVulnetic, vulnetic.ai/enterpriseread 2026-10-06 |
||
|
Testing, Debugging & Optimization Testing, debugging, scoring, retries, fallbacks, quality gates, and optimization loops for improving agent workflows before and after deployment. |
||
|
EnclaveTesting, Debugging & Optimization Triage reproduction and retests after deployment check the customer's systems and fixes. Enclave's Hacking Arena posts benchmark frontier models on hacking tasks, reporting for example that GPT-6 Sol went four for four without faking a single flag. Customers get no harness, scored test cases or quality gate for testing Enclave's agents themselves. SourceEnclave, enclave.airead 2026-10-06 |
||
|
VulneticTesting, Debugging & Optimization Vulnetic tests web applications, APIs, internal networks, cloud and Active Directory, all parts of the customer's estate, and does not evaluate the customer's agents. Its blog promises AI pentesting benchmarks alongside research, but no way to evaluate Sable itself is published. One case reports Domain Admin in 27 minutes. SourceVulnetic, vulnetic.ai, /blog and /active-directory-pentestingread 2026-10-06 |
||
| Specialist automation | ||
|
Browser & Computer Use Browser, desktop, or remote/local computer control for workflows that cannot be handled through stable APIs alone. |
||
|
EnclaveBrowser & Computer Use Live application tests replay requests and record responses. Enclave describes no agent operating a browser or desktop. SourceEnclave, enclave.airead 2026-10-06 |
||
|
VulneticBrowser & Computer Use Sable runs security tools and commands against targets, from authenticated web apps to Active Directory. Vulnetic names no browser or GUI computer use as a capability. SourceVulnetic, vulnetic.ai/ai-pentesting-agent and /web-application-pentestingread 2026-10-06 |
||
Pricing snapshot
Sourced from the Index pricing dataset · open each vendor's profile for full detail.
| Pricing | E Enclave |
V Vulnetic |
|---|---|---|
|
Entry price Lowest public entry point |
Contact sales; no public pricing | Contact sales; pricing is scoped to environment, testing volume and parallel agents. No pricing page. |
|
Pricing confidence How public the numbers are |
Contact only | Contact only |
|
Billing Primary billing axis |
not disclosed | Environment, testing volume and parallel-agent requirements, per the on-premises page. |
|
Variable cost Workload / overage exposure |
Medium variable cost | Medium variable cost |
|
Free tier / trial Try before you buy |
No free tier
|
No free tier
|
|
Buying motion Self-serve vs sales call |
Sales call | Sales call |
|
Pricing verified When this pricing was last checked |
Product updates
Newest first, each linked to its source.
Other matchups in security and SOC agents
Not the pairing you were after? These compare a different set of security and SOC agents on the same 14 capabilities.