Struct SenderCandidate 

struct SenderCandidate {
    name: String,
    tier: &'static str,
    last_seen_ms: Option<i64>,
    stale_eligible: bool,
    channel: Option<String>,
    local_url: Option<String>,
}

One already-fetched candidate, tagged with the tier it was found on. Keeps classify_sender pure and independent of AgentRegistration/ AgentEntry/LanInstance’s actual (differently-shaped) types — the handler below does the field mapping once, at the IO boundary, same discipline as dock_node_views/last_error_frame above.

Fields

name: String

tier: &'static str

"host" | "cross-channel" | "lan" | "wan" — NOT "spawner"; that tier is checked entirely client-side by muxspect.mjs before this route is ever called (it depends on the CALLER’s own environment, which this handler has no way to observe). See docs/specs/SPEC_MUXSPECT_VERIFY_SENDER_2026_08_21.md.

last_seen_ms: Option<i64>

stale_eligible: bool

Whether last_seen_ms is a genuinely-refreshed heartbeat this candidate’s tier can be judged stale by. false for host tier — ReactiveHandler::list_agents() is synchronously accurate (an agent is removed from it on unregister, not aged out by timestamp; see bootstrap/network.rs’s 20s heartbeat task’s own doc comment: “always accurate, with no staleness window of its own”) and nothing in this codebase currently calls update_last_seen to refresh AgentRegistration.last_seen after registration — applying a staleness cutoff to it would eventually flag every healthy, long-running host-tier agent as stale (#2702). false for WAN too (last_seen_ms is always None there — cloud_subscriber doesn’t track a per-agent heartbeat). true for cross-channel (the same 20s heartbeat task above re-writes the shared registry’s updated_at for every live host-tier agent) and LAN (mDNS peers re-announce and bump last_seen on every ServiceResolved).

channel: Option<String>

local_url: Option<String>