formael.com — The agentic outbound gateway. Formael sits at the formal boundary between AI agent intent and external system execution — enforcing policy, abstracting connectivity, and making every action auditable.

Agents speak intent. Formael evaluates it, governs it, executes it, and writes an immutable record of what happened and why.

• Intent Execution Cycle — agents describe what they want; Formael translates intent into authenticated, governed API calls across external SaaS systems
• Four-axis policy engine — identity · semantic · fiscal · risk — versioned, simulatable, always the source of truth
• Human-in-the-loop — high-risk actions pause for human approval before anything crosses the boundary
• Immutable audit ledger — every intent, policy verdict, and execution step recorded together, permanently

AI proposes. The policy engine decides.

actlint — a static action-risk linter for MCP servers.

actlint reads the labels on outbound tool actions and tells you whether they're honest. Deterministic, offline, zero telemetry. No model in the scoring path.

npx actlint <server>

actlint reads the labels on outbound actions. Formael governs the actions themselves.