Enum RoutingMode 

pub enum RoutingMode {
    Observe,
    Cost,
    Balanced,
    Quality,
    Latency,
    Max,
}

Named routing-mode presets: coherent (accuracy/cost/latency) constraint bundles layered on top of the existing escalation knobs. A mode is a resolver over existing config, not a new engine — it overrides only the knobs it has an opinion on; everything else comes from the route/global config unchanged.

Balanced (the default) is a strict no-op: with no mode set anywhere, behaviour is byte-identical to existing behaviour. All other modes are opt-in overlays.

Set per-request via the x-firstpass-mode header, per-route via routing_mode = "...", or globally via the FIRSTPASS_MODE_PROFILE env var.

Variants

Observe

Forward without routing or gating; verdicts are asynchronous learning signals only. Maps to the existing observe passthrough path. Zero added latency, zero quality proof.

Cost

Prefer cheapest start; no speculative prefetch. Tradeoff: lowest token spend; can serve lower quality on hard or heavy traffic.

Balanced

Today’s default — bandit start, configured thresholds, configured speculation. This preset is a strict no-op: byte-identical to existing behaviour when applied.

Quality

One extra escalation rung allowed; serial (no speculative prefetch waste). Tradeoff: higher quality ceiling at higher cost; still bounded by gate verification.

Latency

speculation = 1 always: always prefetch one rung ahead to cut p95 latency. Tradeoff: pays 1 wasted speculative call when the cheap rung passes.

Max

Start at the top (most expensive) ladder rung; verification as insurance. Tradeoff: highest quality, highest cost per call; bandit is bypassed; savings are minimal.

Implementations

impl RoutingMode

pub const ALL: &'static [Self]

All named modes in display order, for discovery surfaces (capabilities, MCP, CLI).

pub fn as_str(self) -> &'static str

Lowercase string name, for use in headers and trace fields.

pub fn preset(self) -> ModePreset

The knob overrides this mode applies on top of the route/global escalation config. None fields are left unchanged — modes only override where they have an opinion.

Balanced returns all None/false: it is a strict no-op by invariant.

Trait Implementations

impl Clone for RoutingMode

fn clone(&self) -> RoutingMode

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Copy for RoutingMode

impl Debug for RoutingMode

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for RoutingMode

fn default() -> RoutingMode

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for RoutingMode

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Eq for RoutingMode

impl PartialEq for RoutingMode

fn eq(&self, other: &RoutingMode) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl Serialize for RoutingMode

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for RoutingMode