Profiles

Profiles let MAPLE keep one runtime model while changing the operating constraints for different environments. A profile shapes which couplings are allowed, which approvals are required, and which safety guarantees dominate.

Common profile set

| Profile | Best fit | Primary concern | | --- | --- | --- | | Human | Human participants in mixed environments | Agency protection | | Agent | General service agents | Controlled tool authority | | Financial | Payment and trading workflows | Audit and bounded consequence | | World | Experiential or simulation contexts | Reversibility and interaction safety | | Coordination | Pure AI-to-AI coordination | Throughput with explicit bounds |

Selection guidance

• Choose Human when disengagement and consent are architectural constraints.
• Choose Agent for general-purpose service automation.
• Choose Financial when approvals, idempotency, and evidence are mandatory.
• Choose World when human-facing experience and reversibility matter.
• Choose Coordination for high-throughput agent meshes with no direct human coupling.

Cross-profile thinking

Profiles are not just labels. They determine what a safe coupling means, what the budget for consequence is, and which Guard rules should apply before execution.

Related reading

• /docs/reference/invariants
• /docs/architecture/worldline-model