MintMCP
by Dependable AI, Inc.
Enterprise MCP gateway that gives every AI agent its own identity, scoped tools and audit trail
MintMCP is an enterprise gateway for the Model Context Protocol that sits inline between AI clients such as Claude, Cursor and ChatGPT and the internal systems they want to reach. It is built for security and platform teams who have lost track of which MCP servers developers are running, and it replaces scattered personal credentials with centrally managed, role-scoped tool access plus a complete audit trail of every tool call.
MintMCP, operated by Dependable AI, Inc. of San Francisco, is an enterprise Model Context Protocol gateway that launched publicly on 5 February 2026. It was founded by Jiquan Ngiam (CEO, previously Google Brain and Coursera), Vijay Vasudevan (previously Google) and Bozhi See (previously Brilliant Home and Shopkick); the founding team includes early Google Brain members who worked on TensorFlow, and the company is advised by Andrew Ng. Backers include Coatue, Maven Ventures, Hustle Fund and WVV Capital, with angels including Andrej Karpathy, Jeff Dean, Scott Belsky and Tom Willerer, though the round size has never been disclosed. The product has two halves. The MCP Gateway hosts a catalogue of pre-approved MCP servers — over 100 managed integrations spanning Snowflake, BigQuery, Databricks, Slack, Microsoft Teams, Google Drive, SharePoint, Salesforce, Linear and Notion — and presents them to users through Virtual MCP Bundles, curated tool sets mapped to organisational roles and kept current through SCIM directory sync, with SSO, SAML and OAuth 2.0 authentication and centrally brokered credentials so developers never hold raw secrets. Agent Monitor is the second half, giving real-time visibility into agent behaviour by capturing file operations, bash commands and API calls and applying rule-based detection and blocking for PII exposure, credential leakage, risky commands and prompt-injection attempts. Everything traverses the gateway as an inline proxy, so every request is authenticated, routed, policy-checked and logged, with audit retention configurable from 90 days to seven years and SIEM export to Microsoft Sentinel, Splunk or S3. The company is SOC 2 Type II audited and offers a HIPAA BAA. Deployment defaults to managed SaaS with self-hosting available through an enterprise engagement, and by September 2026 MintMCP reported more than 40 paying customers including Braze, Coursera, Stability AI, Modern Treasury, Cursor and Arlo, on a team of roughly a dozen people.
The security or platform engineering lead at a company where developers have already wired personal MCP servers into Claude, Cursor or ChatGPT and nobody can answer which systems those agents can currently reach.
One inline chokepoint where every agent tool call is authenticated against a role-scoped bundle, inspected for risky behaviour and written to an exportable audit log.
At a Glance
- Category
- Governance & Security
- Pricing
- Subscription, Contact for pricing
- Target Market
- CISOs, CTOs, Platform Engineers, Security Engineers, Enterprise Developers
- Deployment
- Cloud-first, Self-hosted
- Headquarters
- San Francisco, United States
- Team Size
- 11-50
- Customers
- 40+ paying customers as of September 2026, including Braze, Coursera, Stability AI, Modern Treasury, Cursor and Arlo
Key Features
- ✓Managed MCP Gateway
Hosts and lifecycle-manages 100+ pre-approved MCP integrations so teams stop running connector infrastructure themselves.
- ✓Virtual MCP Bundles
Curated tool sets mapped to organisational roles, kept in sync automatically through SCIM directory provisioning.
- ✓Agent Monitor
Captures file operations, bash commands and API calls in real time, detecting and blocking PII exposure, credential leakage and prompt injection.
- ✓Centralised credential brokerage
Pre-configured credentials are held at the gateway so users connect after authenticating without ever holding raw secrets.
- ✓Audit logging and SIEM export
Complete tool-call audit trails retained from 90 days to seven years and exportable to Sentinel, Splunk or S3.
- ✓Enterprise identity integration
SSO, SAML, OAuth 2.0 and SCIM give agents identities managed separately from the human credentials they previously borrowed.
Capabilities
Use Cases
- •Shadow MCP cleanup
Inventory the MCP servers developers wired up themselves and replace them with an approved central catalogue.
- •Coding agent guardrails
Watch what Claude Code or Cursor actually does on a repository and block risky bash commands before they execute.
- •Regulated data access
Let analysts query Snowflake or BigQuery through an agent while every request is policy-checked and permanently logged.
- •Offboarding and access review
Revoke an employee's agent tool access centrally through SCIM instead of hunting for credentials on individual machines.
- •Security incident forensics
Export the full tool-call trail to Splunk or Sentinel to reconstruct precisely what an agent touched during an incident.
Ideal For
Best For
- ✓Replacing developer-managed MCP server configs with a centrally approved, role-scoped catalogue
- ✓Removing raw credentials from developer laptops by brokering them at the gateway instead
- ✓Producing an auditable record of every agent tool call for security and compliance review
- ✓Detecting and blocking risky coding-agent behaviour such as credential leakage or prompt injection
- ✓Giving regulated teams an MCP path that already carries SOC 2 Type II and a HIPAA BAA
Not Ideal For
- ✗Latency-sensitive agent workloads — independent gateway comparisons measure 100-250ms of added overhead from MintMCP's deep inspection and compliance scanning
- ✗Organisations that will not let a third party broker credentials, since the default deployment is managed SaaS and self-hosting is available only on enterprise request
- ✗Buyers who need a mature vendor with a public track record; the company launched in February 2026, has roughly a dozen staff and has never disclosed its funding round
- ✗Teams wanting to evaluate self-serve on a published price, because every deal is custom-quoted through sales
Deployment
Market & Ratings
40+ paying customers as of September 2026, including Braze, Coursera, Stability AI, Modern Treasury, Cursor and Arlo
Market Analysis
Pros
- ✓SOC 2 Type II audited with a HIPAA BAA available, unusual for a company that only launched publicly in February 2026
- ✓Managed hosting of 100+ pre-approved MCP integrations removes the connector infrastructure work most teams currently do by hand
- ✓Agent Monitor adds behavioural detection — PII exposure, credential leakage, risky bash commands, prompt injection — on top of plain gateway routing
- ✓Audit log retention is configurable from 90 days to seven years with SIEM export to Sentinel, Splunk or S3
- ✓Named customers are real and checkable, including Braze, Coursera, Modern Treasury, Arlo and Cursor
Cons
- ✗Independent gateway comparisons measure 100-250ms of added latency from MintMCP's deep inspection and compliance scanning, acceptable for most enterprise work but not for latency-sensitive agents
- ✗The gateway enforces policy at the routing layer but still trusts the connected MCP servers, so a misconfigured or over-privileged backend server remains the weak link
- ✗The managed-SaaS default concentrates credential brokerage in a third party, which some regulated buyers will refuse; self-hosting is only available by enterprise request
- ✗Parts of the guardrail stack are delegated to third parties such as Microsoft Purview, Google Cloud DLP and Amazon Bedrock rather than built in-house
- ✗No G2, Capterra or TrustRadius listing and essentially no Reddit or Hacker News discussion beyond the founders' own posts, so there is no independent practitioner track record
- ✗The team was roughly twelve people in September 2026 and the funding round has never been disclosed, so runway and support depth are unknowable from outside
Pricing
Custom / Enterprise
Contact for pricing
- ✓MCP Gateway with managed MCP server hosting
- ✓LLM Proxy to route and monitor model traffic
- ✓Virtual MCP Bundles with role-based access
- ✓SSO/SAML and SCIM directory sync
- ✓OpenTelemetry export and audit logging
- ✓SLAs, high availability and deployment assistance
No list price is published. MintMCP's own pricing page describes custom deals as per-user licensing plus a scalable platform fee, sized by team and requirements, with enterprise add-ons covering SSO/SAML, SCIM directory sync, OpenTelemetry export, dedicated success management and SLAs. Self-hosting is available only through an enterprise conversation rather than self-serve, so even a proof of concept means talking to sales first.
Security & Compliance
Sources
This page was written from 7 sources, 3 on domains other than mintmcp.com.
Stay Ahead of the Curve
Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.
SubscribeRelated Products
JFrog AI Catalog
Govern the models, MCP servers, skills and plugins your AI coding agents consume
Alice (formerly ActiveFence)
AI red teaming and real-time guardrails for models, apps and agents, built on a decade of adversarial data
Geordie AI
Discover, govern and cost-account every AI agent running across your enterprise
CrowdStrike Falcon Guardian
AI detection and response that finds shadow AI agents on the endpoint and blocks the unapproved ones