What is C1 MCP Gateway?

C1 MCP Gateway governs selected agent, human, and workload actions on supported MCP, tool, and API targets through a mediated path or direct Cross-App Access.

Does C1 MCP Gateway route LLM requests?

No. C1 MCP Gateway governs tool and API actions. C1 LLM Gateway handles policy-based routing across supported inference providers and model deployments.

What is the difference between a proxy path and direct Cross-App Access?

On a proxy path, C1.ai mediates a supported tool or API call and applies inline policy. On a supported direct Cross-App Access path, C1.ai issues a short-lived scoped token after policy and the client calls the target directly, so application payload does not traverse C1.ai.

Can C1 MCP Gateway inspect every direct target call?

No. Direct-path evidence begins with the C1.ai policy decision and token issuance. Tool-level request, response, and outcome evidence requires an explicit target event contract.

How are sensitive actions handled?

Target policy can allow lower-risk actions, hold sensitive actions for an accountable owner or approver, and deny destructive actions. The exact policy and supported controls depend on the selected target path.

Product

C1 MCP Gateway

Put identity-aware policy in front of every MCP tool call and resource access.

Book a demo

3000+ out of the box MCPs with identity-aware access controls.

Connect agents to every approved tool

Give agents and AI clients one governed path to approved MCP servers, SaaS, APIs, databases, and on-prem systems.

Enforce policy on every action

Evaluate identity, current access, action risk, and company policy on every call. Auto-approve low-risk work, require approval for higher-risk actions, and deny anything outside policy.

Capture every decision and outcome

Record the caller, tool, requested action, policy decision, approval, and outcome for investigations, access reviews, and audits.

A unified gateway for MCP tool management and Agent SSO (XAA)

TOOL MANAGEMENT

New AI tools, governed in minutes.

How it works

Authenticate to C1.ai once, and it brokers access to every tool downstream: a single login where Cross-App Access is supported, centralized auth everywhere else.

ROLE BASED ACCESS

Assign access from agent identity and role.

How it works

When an agent acts for a person, start from that person's C1.ai identity and current grants so the agent never inherits broader access than its user.

Assign access from agent identity and role.

GUARDRAILS

Govern the action, on every call.

How it works

Every tool ships graded by risk. Reads auto-approve, writes route to an approver, and destructive calls are denied by default. Every grade is set by policy, so you tune it to your risk.

PERMISSION ESCALATION

Agents escalate permissions requests seamlessly

How it works

Ask for an AI tool in Slack, Teams, the CLI, or your IDE, and policy grants it in seconds. No ticket, no wait.

Agents escalate permissions requests seamlessly

SELF SERVICE MCP ACCESS

Make AI tools and MCP access self-service

How it works

Make MCP servers, toolsets, and individual tools discoverable without exposing anything that has not passed administrator review.

Make AI tools and MCP access self-service

AUTHN & AUTHZ

Authenticate every agent. Authorize every action.

How it works

Let an agent act in a person's authorized context so every tool call starts from that user's current identity, roles, and grants.

Authenticate every agent. Authorize every action.

C1.ai solutions that meet your needs

Agent SSO & Runtime GovernanceRun agents securely, without standing credentials or unchecked access.Deploy Custom AgentsTurn a defined goal into governed, multi-step automation across enterprise systems.Apps & MCP Self ServiceGive on-demand access to approved apps, MCPs, and AI tools.Shadow AI DiscoveryFind unsanctioned AI tools, agents, MCP servers, and credentials.

Connect every agent to enterprise tools with identity, policy, and guardrails.

Book a demo

C1 MCP Gateway

3000+ out of the box MCPs with identity-aware access controls.

Connect agents to every approved tool

Enforce policy on every action

Capture every decision and outcome

A unified gateway for MCP tool management and Agent SSO (XAA)

New AI tools, governed in minutes.

One place to authenticate

Scale tools without scaling cost

Reach on-prem, no firewall changes

Connect to everything you run

Bring your own AI clients

Assign access from agent identity and role.

Delegate access from a human user

Give standalone agents their own identity

Authenticate workloads through federation

Assign standing or time-bound roles

Recheck access on every tool call

Govern the action, on every call.

Right-size every approval

Blocks the attacks you can't see

Safe even when it breaks

Agents never hold your secrets

Cut off a compromised agent

Agents escalate permissions requests seamlessly

No tickets, no waiting

Agents ask for what they need

Revoke it, and the next call stops

Backed by real identity governance

Make AI tools and MCP access self-service

Publish approved tools in the access catalog

Request access where work already happens

Apply policy before granting access

Grant only the tools and duration required

Remove access when the need ends

Authenticate every agent. Authorize every action.

Delegate authorization from a human user

Give standalone agents a governed identity

Use Agent SSO where XAA is supported

Enforce every proxied call through C1.ai Gateway

Keep downstream credentials out of user and agent hands

Bind the connection with DPoP

C1.ai solutions that meet your needs

Connect every agent to enterprise tools with identity, policy, and guardrails.

FAQs

What is C1 MCP Gateway?

Does C1 MCP Gateway route LLM requests?

What is the difference between a proxy path and direct Cross-App Access?

Can C1 MCP Gateway inspect every direct target call?

How are sensitive actions handled?

A unified gateway for MCP tool management and Agent SSO (XAA)

New AI tools, governed in minutes.

One place to authenticate

Scale tools without scaling cost

Reach on-prem, no firewall changes

Connect to everything you run

Bring your own AI clients

Assign access from agent identity and role.

Delegate access from a human user

Give standalone agents their own identity

Authenticate workloads through federation

Assign standing or time-bound roles

Recheck access on every tool call

Govern the action, on every call.

Right-size every approval

Blocks the attacks you can't see

Safe even when it breaks

Agents never hold your secrets

Cut off a compromised agent

Agents escalate permissions requests seamlessly

No tickets, no waiting

Agents ask for what they need

Revoke it, and the next call stops

Backed by real identity governance

Make AI tools and MCP access self-service

Publish approved tools in the access catalog

Request access where work already happens

Apply policy before granting access

Grant only the tools and duration required

Remove access when the need ends