Model Context Protocol
Yes. Agents discover and invoke MCP tools only when policy grants them, and every call is stored on the execution trail.
Open protocol for tool and resource access—treated as part of Obliq’s execution system.
Who MCP on Obliq is for
Teams that already expose tools as MCP servers, or that want one protocol instead of a custom adapter per agent framework. Obliq does not replace the server you run; it is the client, the policy, and the record of the call.
How an MCP call works
Register the server. Grant specific tools to an agent. When the agent selects one, Obliq checks the schema, optionally waits for confirmation, invokes the tool, and writes the arguments and result onto the execution next to retrieval and inference.
What MCP provides
MCP standardizes how AI systems discover tools and resources. In Obliq, those servers sit on the same tool bus as REST APIs, webhooks, and functions—scoped by agent policy, schema-validated, and fully traced.
{
"protocol": "MCP",
"tool": "zendesk_create_ticket",
"arguments": {
"priority": "high",
"client_id": "ACME-01",
"sla": "99.99%"
}
}- 01 Agents only receive tools they are explicitly granted.
- 02 Inputs must match registered JSON schemas.
- 03 Sensitive actions can require human confirmation before firing.
- 04 Every call is stored on the execution ledger for audit and replay.
Questions
What is the Model Context Protocol?
MCP is an open protocol for discovering and invoking tools and resources from an AI system. Obliq treats MCP servers as first-class connectors on the tool bus.
Does Obliq support MCP?
Yes. Obliq includes native MCP client and server support. Agents can discover and invoke granted tools, and each invocation is recorded for audit and replay.
How does Obliq secure MCP tool calls?
Agents only receive tools they are explicitly granted. Inputs must match registered JSON schemas. Sensitive actions can require human confirmation. Every call is stored on the execution ledger.