Skip to main content
Interf provides the open Capabilities Protocol and Runtime infrastructure to solve the Forward Deployed Engineer (FDE) Crisis. We treat Agent Onboarding as Code: requirements are declared in a machine-readable contract, then evaluated against the enterprise environment to produce a deterministic readiness state.

​
The Problem: Blind Agents

Agents are designed to be autonomous, but today they are deployed “blindly” into opaque enterprise environments. They do not know if they have the necessary Data Access, Authentication, or Compliance permissions to function. This leads to a structural failure where readiness gaps are discovered mid-flight. Organizations are forced to rely on human FDEs to manually patch these gaps, causing massive stalls and frozen capital.

​
The Solution: Protocol + Runtime

Interf sits between Agent Logic and Enterprise Context.

The Protocol

Agentic Capabilities Protocol (ACP) A machine-readable standard for declaring Agent requirements (data, auth, compliance) as a Capabilities Contract.

The Runtime

Interf Runtime (MCP Server) Infrastructure that evaluates the contract against the enterprise environment and exposes a readiness state that agents can query to self-onboard.

​
The Mechanism

  1. Declare: Vendors define requirements in a capabilities.yaml contract.
  2. Evaluate: The Interf Runtime checks the contract against local resources (Data, Auth, Rules).
  3. Onboard: The Agent connects via Model Context Protocol (MCP) to discover its state and self-onboard.
# capabilities.yaml (The Contract)
agent: "financial_analyst_agent"
version: "0.1.0"

capabilities:
  - id: "ledger_write_access"
    description: "Allows agent to update Q4 forecasts"
    prerequisites:
      - type: "data"
        resource: "snowflake_finance"
        access: "write"
      - type: "compliance"
        standard: "soc2_type_ii"
Status: The Protocol Specification and Runtime (v0.1) will be published here in January 2026.