Tenable OPEN for AI Bets on Openness in a Security Market Built on Lock-In
The exposure management vendor is wiring its data into partner and customer AI tools over MCP, with a marketplace for agent skills. The pitch is choice; the question is whether openness sells.
Tenable has expanded its open technology and partner ecosystem with a program called Tenable OPEN for AI, which connects its exposure management data to the AI tools its customers already use. Announced on 20 October 2026, the program links Tenable’s Hexa AI MCP server with partner MCP servers, so that security teams can interrogate data across several vendors from one AI chat window. It launched with twelve partners, among them SentinelOne, Splunk, Snyk, BeyondTrust, Corelight, Infoblox, Securonix, Recorded Future and Cohesity, and with four use cases available immediately.
The company deliberately positions its pitch against vendor lock-in. Tenable’s chief product officer, Eric Doerr, describes the company as the only exposure management provider offering a truly unrestricted, vendor-agnostic ecosystem of validated AI use cases. That is a positioning statement as much as a product one, and it is worth testing: Tenable is a public company with a platform to sell, and openness is a strategy, not an ideology. What makes the launch interesting is that it is a concrete bet on a specific technical choice, the Model Context Protocol, as the way security tools will talk to agents.
What Tenable launched
Tenable OPEN for AI sits inside the Open Partner Exchange Network, the company’s technology partner program, and delivers validated, ready-to-use AI workflows. Each use case is available in two forms: as a documented workflow that a security team can replicate with the AI tool of its choice through the Tenable Hexa AI MCP server, or as a ready-to-use skill on the CyberAgents Exchange, the vendor-agnostic AI agent marketplace Tenable launched in August 2026. The launch partners are BeyondTrust, Cohesity, Command Zero, Corelight, Infoblox, Recorded Future, Securonix, SentinelOne, Setyl, Snyk, Sparkle In and Splunk. Four featured use cases are available now, and the program builds on the Tenable One platform and its Hexa AI agentic engine, which Tenable announced in March 2026.
The marketplace is not new. Tenable launched the CyberAgents Exchange in August 2026 at Black Hat, describing it as the industry’s first open-source AI agent exchange and a vendor-agnostic registry for agents, skills, MCP servers and multi-agent playbooks, with SentinelOne and Recorded Future as founding members and more than 50 components released under open-source licences.
The Exchange is free to use and does not require a Tenable customer relationship, and listings link to code hosted in contributors’ own repositories rather than acting as a code host. OPEN for AI effectively feeds validated, commercially supported use cases into that community directory, which is why the program reads as an ecosystem play rather than a single product.
The mechanism: MCP as the integration layer
The technical bet is straightforward. Rather than build bespoke integrations between each pair of security products, Tenable is using the Model Context Protocol, an open standard that lets AI applications connect to external tools and data sources through a common interface. By connecting the Tenable Hexa AI MCP server with a partner’s MCP server, a customer can ask a question in one AI client and have the answer assembled from several systems at once, with no custom development.
The examples Tenable gives are concrete: joining a Snyk code finding with Tenable runtime data to see whether a vulnerability is actually deployed, or combining BeyondTrust identity data with Tenable findings to work out which privileged accounts can reach a vulnerable host. Tenable’s position in the market is relevant to how the bet should be read. It is among the largest pure-play exposure management vendors, with more than 40,000 customers and a platform, Tenable One, spanning vulnerability management, cloud, identity and operational technology. It does not sell the endpoint agent or the SIEM that several of its new partners do, which makes it a natural connective layer: it is more useful to Splunk, SentinelOne and Snyk as a data source than it is a direct competitor to them.
The openness, in other words, is not altruism; it is a strategy that fits a vendor whose value lies in the breadth of its exposure data rather than the breadth of its product surface. This is the same pattern appearing across the industry. Security vendors from Wiz to CrowdStrike and Rapid7 now ship MCP servers of their own, and MCP is becoming the default wiring between agents and enterprise systems.
The bet on openness
The strategic question is why a company would open its data to rivals’ agents rather than keep it inside its own console. The answer is that the value of exposure data grows with how many places it can be used, and customers increasingly refuse to standardise on one vendor’s AI. Tenable’s argument is that a security team will pick the tool that gives it the most freedom, and that being the connective layer across the stack is worth more than the lock-in it gives up.
There is a counter-argument worth stating: openness can commoditise the vendor whose data flows into someone else’s interface, and Tenable is giving partners a way to surface its findings inside their own workflows. The company’s answer is the marketplace, the CyberAgents Exchange, which it positions as vendor-neutral and free, and which gives it a community role beyond its own product. Whether that is a moat or a giveaway is the question the next year will answer.
There is a structural reason the question matters. Security buyers have spent a decade consolidating onto platforms, and platform vendors have spent the same decade arguing that integration is a feature of scale. The rise of agentic workflows threatens that logic, because an agent can join data across products without the products being consolidated at all. If that holds, the advantage shifts from owning the most modules to owning the most useful data and the best-supported workflows, which is precisely the ground Tenable is staking out.
The use cases
The four featured use cases illustrate where the company intends to deliver value. Cross-Platform Security Triage, with SentinelOne, lets an analyst investigate alerts and misconfigurations in natural language from one chat window. Network Threat and Blast Radius Mapping, with Corelight, joins network telemetry to Tenable asset context to prioritise by potential blast radius.
Code-to-Runtime Risk Correlation, with Snyk, checks whether a code-level vulnerability is actually running in production. Identity-Enriched Exposure Triage, with BeyondTrust, risk-ranks assets by which identities can reach them. Each is a join that was previously manual, and each is a small, concrete answer to the question every exposure finding raises: does it matter, and who can exploit it?
The market it enters
Tenable competes in exposure management against CrowdStrike, Rapid7, Qualys, Microsoft, Wiz and others, and its rivals have been shipping MCP servers too, so openness is not unique to Tenable. What is distinctive is the breadth of the partner list and the marketplace. There is also an irony worth noting: MCP servers are themselves a growing security problem.
Researchers and vendors have flagged that many MCP servers run with static secrets, lack authentication and expose more than they should, and Qualys has built product features to discover them as shadow IT. A program that wires dozens of MCP servers into security workflows is, therefore, both a productivity play and an expansion of the attack surface it exists to manage.
What buyers should ask
For a security leader weighing it, the questions are practical. Which of our existing tools have MCP servers, and which of Tenable’s validated use cases apply to our stack? How do you enforce permissions when an agent reads from several systems in one session to prevent it from surfacing data a user should not see? What do you log, and how can we audit each agent’s actions? How do you vet partner MCP servers, and how do you assess their security posture?
Does this reduce the number of consoles our analysts touch, or add another layer to govern? And is the openness durable, or a go-to-market phase that could change if the economics shift? It is also worth asking how the arrangement changes if a partner later becomes a direct competitor, since several of them sell overlapping products.
What it signals
Tenable OPEN for AI is a sensible, well-timed move from a vendor whose data is more useful the more places it can be used. It also marks a shift in how security tooling is assembled: the integration layer is becoming the agent protocol, and vendors that ship an MCP server and a use-case library are competing to be the connective tissue. The honest caveats are the familiar ones. Tenable and its partners validate the use cases rather than independent parties, the company uses openness as a commercial strategy that it could change over time, and MCP remains young and lacks robust security.
The program’s real test is whether security teams adopt the workflows in production, and whether openness wins them from platforms that would rather keep the data inside. That is why the launch is worth watching beyond Tenable’s own numbers.
If open, MCP-based ecosystems win adoption, the security stack of the next few years will look less like a platform and more like a set of interoperating tools joined by agent protocols, and the vendors that made their data easy to reach will be the ones agents reach for. If they do not, consolidation resumes and the openness was a phase. Tenable has placed a visible bet on the first outcome.