Stacklok
Also known as: ToolHive, Stacklok Enterprise, vMCP
Self-hosted platform for running and governing what agents reach: MCP servers in isolated containers, an identity-aware tool gateway and an AI gateway for model routing, budgets, screening and audit, on Kubernetes.
Stacklok gives enterprises a self-hosted way to run and govern the tools and models their AI agents use, on the Kubernetes infrastructure they already trust. Stacklok says the co-creators of Kubernetes, Craig McLuckie and Joe Beda, are building it. It is built on ToolHive, an Apache 2.0 open source platform that runs each MCP server in an isolated container with its own permissions and secrets, available as a desktop app, CLI and Kubernetes operator, with a Registry Server for MCP servers, agent skills and plugins and a Virtual MCP Server that aggregates several backends behind one endpoint with tool filtering.
Stacklok Enterprise adds two gateways that bookend an agent. The Connector Gateway, which Stacklok also calls its MCP Gateway, is an identity-aware endpoint for tool calls: administrators register MCP servers as connectors, grant them to directory groups and let the gateway broker credentials, so every call carries the identity of the person behind it, and a usage view shows traffic by connector and tool.
The AI Gateway is the same idea for models: connect OpenAI, Anthropic, Bedrock, Azure OpenAI, Vertex AI or any OpenAI-compatible provider, route model names with weighting, failover and retries, cap spend per user or group in dollars, screen prompts for injection and scan for card and personal data, and forward structured audit events to a SIEM. An Enterprise Manager enforces policies across clients, and authorization is expressed as Kubernetes custom resources mapping identity-provider groups to MCP roles.
Everything runs in the customer's own cluster, installed as one Helm chart, including air-gapped, and the company neither hosts nor processes customer data; the distribution carries Sigstore signatures, SLSA provenance and SBOMs. ToolHive and the gateways govern what agents may reach and record what they did; they do not run agent workflows, hold agent memory or index the customer's documents. Mecatl, a separate open source agent harness from Stacklok with its own site at mecatl.dev, runs the agents themselves. ToolHive is free; Stacklok Enterprise is an annual subscription quoted through sales.
Vendor details
Canonical URL
https://stacklok.com
Category
Agent infrastructure
Subcategory
Enterprise MCP runtime and governance
Funding status
Independent; per its homepage, backed by Accel, Bain Capital Ventures and Madrona, led by CEO Craig McLuckie and CTO Joe Beda, and a charter member of the Agentic AI Foundation.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
Runs any MCP server in an isolated container, with a curated registry of vetted servers and agent skills, and a Connector Gateway that brokers each connector's credentials per user. Identity comes through Okta and Microsoft Entra ID with SCIM provisioning. Model providers OpenAI, Anthropic, AWS Bedrock, Azure OpenAI, Google Vertex AI and OpenAI-compatible endpoints connect through the AI Gateway. OpenTelemetry and SIEM forwarding carry traces and audit events. Delivery spans a Kubernetes operator, CLI, desktop app and GitHub Actions.
In practice
A platform team runs vetted MCP servers through the ToolHive Kubernetes operator in its own cluster, each in an isolated container with its own permissions and secrets.
A security team maps Okta or Microsoft Entra ID groups to MCP roles, traces every tool call through OpenTelemetry into its own stack and forwards audit events to its SIEM.
An enterprise curates a registry of approved MCP servers, and Enterprise Manager policies block any server outside it, so developers install from an internal catalog instead of random GitHub repos.
Sources & related URLs
Agentic Index coverage score
8.0 / 14 capabilities · 57%
| Integrations & Tool Calling | Full |
|---|---|
|
Every tool call carries the identity of the person who made it. Administrators register MCP servers as connectors in the Connector Gateway, an identity aware endpoint for MCP tool calls, and grant them to directory groups. The gateway then brokers each connector's credentials on the caller's behalf. ToolHive runs each MCP server in an isolated container with its own permissions and secrets, and a Virtual MCP Server aggregates backends into one endpoint with tool filtering for a given agent. Sourcedocs.stacklok.com/connector-gatewayread 2026-11-03 |
|
| Workflow Orchestration | Not documented |
|
A Virtual MCP Server aggregates backend MCP servers behind one endpoint with static backend selection, tool filtering and renaming. The Connector Gateway exposes a toolset scoped to a group. Both route a call to the right server instead of running work. The platform runs no workflow or agent loop of its own, and there is no sequencing, branching or retrying across deterministic workflow steps and autonomous agent steps. Sourcedocs.stacklok.com/connector-gatewayread 2026-11-03 |
|
| Knowledge Grounding & RAG | Not documented |
|
The vMCP optimizer searches tool descriptions to surface only the relevant tools for each request and cut token use. The registry catalogs MCP servers and skills. There is no index, graph or embeddings layer over the customer's own documents. Sourcedocs.stacklok.com/toolhive/guides-vmcp/optimizerread 2026-11-03 |
|
| Human Oversight & Guardrails | Partial |
|
Administrators limit what an agent may do instead of putting a person in the loop. They decide which MCP servers are registered as connectors and which directory groups may use them. Enterprise Manager policies can block servers that are not in the approved registry, and authorization policies limit each role's tools. The AI Gateway screens prompts for injection and scans requests and responses for card and personal data. The customer controls all of these limits, but no person reviews, approves or validates an agent action before it commits. Sourcedocs.stacklok.com/connector-gatewayread 2026-11-03 |
|
| Security, Identity & Governance | Full |
|
Identity comes from Okta or Microsoft Entra ID, with SCIM provisioning into a directory of users, groups and subgroups. Virtual API keys are bound to a user's identity. Enterprise authorization uses fleet wide MCP roles in RBAC custom resources, with delegation at the namespace level, and Cedar policies apply to MCP servers. Connector credentials are stored encrypted at rest in the directory. The distribution itself carries Sigstore signatures, SLSA provenance and SBOMs the customer can verify. Stacklok holds no third party attestation of its own operations and says none applies to a self hosted product that holds no customer data. Sourcestacklok.com/security-centerread 2026-11-03 |
|
| Observability & Auditability | Full |
|
The AI Gateway records structured audit events for model requests, detections and key actions and forwards them to the customer's SIEM. The Connector Gateway's tool usage view shows traffic by time range, connector and tool. An Enterprise Manager telemetry policy enforces OpenTelemetry settings across Stacklok clients, so tool calls are traced into the customer's own stack. A customer can inspect what an agent sent and received, keep audit events apart from traces, and export both to its own tools. Sourcedocs.stacklok.com/ai-gateway/forward-audit-logsread 2026-11-03 |
|
| Memory & State Persistence | Not documented |
|
The product runs and governs MCP servers, skills and model traffic. None of its components (the ToolHive runtime, Registry Server, vMCP, Connector Gateway, AI Gateway and Enterprise Manager) persists an agent's state or a user's context across runs, so agents get no memory from it. Sourcedocs.stacklok.com/toolhiveread 2026-11-03 |
|
| Deployment & Data Residency | Full |
|
Everything runs inside the customer's own infrastructure. Stacklok Enterprise deploys as a single umbrella Helm chart into the customer's Kubernetes cluster. Distributed deployments can span several clusters, and an air gapped install mirrors the chart from a private registry. Stacklok neither hosts nor processes customer data. Sourcedocs.stacklok.com/platformread 2026-11-03 |
|
| Prebuilt Agents, Templates & Packs | Full |
|
Agent skills ship with ToolHive. They are reusable bundles of tools, permissions and secrets that a customer browses in a registry and installs into its AI clients. A skill can be built from a local folder as an OCI artifact, then published, listed or deleted through the Registry Server's admin API. The CLI and desktop UI also install, distribute and manage skills. Sourcedocs.stacklok.com/toolhive/guides-registryread 2026-11-03 |
|
| Triggers & Channel Coverage | Not documented |
|
A client or agent calling a tool or a model is what sets the platform to work. ToolHive runs the servers and the gateways broker the calls, both at the caller's request. No schedule, event, webhook or inbound queue starts an agent run, and nothing wakes an agent unless a person or client starts the work. Sourcedocs.stacklok.com/toolhiveread 2026-11-03 |
|
| Model Flexibility & Routing | Full |
|
Customers and admins decide which model serves a request. The AI Gateway is a self hosted control point between AI clients and model providers. The customer connects OpenAI, Anthropic, AWS Bedrock, Azure OpenAI, Google Vertex AI and OpenAI compatible providers. It then maps model names onto them with a default route, weighted load balancing, priority failover, retries and timeouts, declared as AIGateway and AIPolicy resources. Sourcedocs.stacklok.com/ai-gatewayread 2026-11-03 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
Builders get ToolHive, an Apache 2.0 open source platform with a CLI, desktop UI and Kubernetes operator. Its Registry Server implements the MCP registry API for servers, skills and plugins. Stacklok Enterprise publishes a REST API reference for the Enterprise Manager, Connector Gateway and AI Gateway. It also publishes custom resource definitions for authorization and a management API for budgets. Sourcedocs.stacklok.com/toolhiveread 2026-11-03 |
|
| Testing, Debugging & Optimization | Partial |
|
Before an MCP server goes to production, ToolHive's CLI can test and validate it. The registry applies review criteria before a server is listed. There are no scored test cases, pass rates or judge verdicts that give a readable result to compare one change against another. Sourcedocs.stacklok.com/toolhive/guides-cli/test-mcp-serversread 2026-11-03 |
|
| Browser & Computer Use | Not documented |
|
The product governs and runs MCP servers and model traffic, and drives no interface itself. It operates no browser, desktop or remote computer session for an agent, and no headless fetch or third party browsing engine is wired in. Sourcedocs.stacklok.com/toolhiveread 2026-11-03 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Product updates
ToolHive v0.57.0 adds authorization scoped to each backend in its virtual MCP server, so access rules can differ for each MCP server sitting behind the gateway. Token exchange now accepts authorization details, and upstream subject identifiers can be switched on as an option. Applications that embed ToolHive can also refuse refresh grants when the upstream credential has stopped working.
Bears on: MCP / tool calling / API
View sourceToolHive v0.53.0 adds group based upstream selection to the embedded auth server of its Virtual MCP Server. Operators can set a filter that picks which additional identity providers each user is sent through, based on group claims from the first upstream login. The feature is opt in and fails closed when the group claim is missing.
Bears on: MCP / tool calling / API
View sourceStacklok has introduced an integration with Anthropic MCP Tunnels to securely connect Claude to internal systems behind corporate firewalls. In this architecture, Stacklok serves as the critical Model Context Protocol management layer that complements Anthropic's tunneling capabilities.
Bears on: Integrations
View sourcePricing
Open source ToolHive free; Stacklok Enterprise quoted through sales
Annual Stacklok Enterprise subscription; open source core free
What is public
What each edition includes is public: the docs publish a Community versus Enterprise capability comparison. No price is published anywhere on the estate.
Billing mechanics
Free open source core; Stacklok Enterprise licensed as an annual subscription quoted through sales, deployed by the customer into its own cluster.
Cost watchouts
The governance layer is the paid part: the AI Gateway, Connector Gateway, Enterprise Manager, SCIM and directory services, the versioned and attested distribution and SLA support are Enterprise. Because it is self-hosted, the customer also runs and pays for the Kubernetes infrastructure the platform and every MCP server sit on.
Variable cost rationale
Enterprise is an annual subscription rather than usage metered, and the free open source core absorbs experimentation, so cost is predictable within a subscription term; connectors and services are the main additive levers.
Additional watchouts
Evaluate on the free open source core, but production governance features (IdP, hardened images, audit, gateways) require the paid annual subscription.
Sales call required
Yes, required for paid access
Free / trial
ToolHive is free and open source (Apache 2.0); no Enterprise trial is published
Key ambiguities
No figure, unit or tier structure is published for Stacklok Enterprise; stacklok.com/pricing does not exist and the docs route to a demo.
Missing data
Enterprise subscription rates, connector licensing rates, services rates.
Related vendors
- AgentOps — Agent observability and debugging platform: open source SDKs trace…
- AgentPhone — AgentPhone gives the AI agents a team builds their own US or…
- Agno — Python agent framework and AgentOS runtime (formerly Phidata) for…
- AIsa — Resource and payment gateway for AI agents, with one key to 110+…
- Albato — Albato is a no code integration platform that sells an embedded,…
- AlphaBitCore — AI control plane for regulated financial firms: one gateway enforces…
Alternatives to Stacklok
The closest documented capability profiles to Stacklok among agent infrastructure platforms tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Arcade7.5 / 14Fuller documented coverage on Testing, Debugging & OptimizationStacklok vs Arcade →
- Portkey7.5 / 14Fuller documented coverage on Testing, Debugging & Optimization
- AIsa6.0 / 14A lighter documented profile than Stacklok
- Arize AI9.0 / 14Adds documented Triggers & Channel Coverage
- Metorial8.0 / 14Adds documented Triggers & Channel Coverage
- Clawvisor6.5 / 14Fuller documented coverage on Human Oversight & Guardrails
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded