Quote before execution
No silent commission. A future paid rail must expose its price before the caller elects to execute.
PANGI is HumanMirror's performance-routing layer for autonomous agents. Today it ranks explicit route measurements, produces a deterministic Proof-of-Latency digest and returns a failover order. The distributed routing mesh is the target architecture—not a deployment claim.
The Ghost Toll is the economic target: a microscopic, explicitly quoted routing fee that only makes sense when the performance or resilience gain is worth more than the marginal cost. It is not a hidden deduction and it is not currently collected by PANGI v1.
No silent commission. A future paid rail must expose its price before the caller elects to execute.
The intended settlement path is compatible with HumanMirror's x402 and non-custodial infrastructure.
The economic rationale is performance efficiency, not forced participation. If the gain is not measurable, the toll has no justification.
Agents submit candidate HTTPS routes plus observed timing and health data. PANGI normalizes the evidence, ranks healthy routes, selects a primary path and emits deterministic fallbacks with a reproducible proof identifier.
The architecture separates what is executable today from what HumanMirror is building toward, so agents can integrate immediately without relying on fictional infrastructure.
Unavailable or invalid routes are excluded from the healthy primary set. Remaining candidates receive a deterministic order.
Geographically distributed nodes, independently observed latency and automatic rerouting are the target network layer.
Execution, explicit fee quotation, settlement and receipt are designed to converge without introducing custodial dependence.
PANGI is discoverable through HumanMirror, OmniDome, the official OmniDome MCP, A2A metadata, llms/agents surfaces and its dedicated well-known manifest.