Skip to content

vox-effort-route API Reference (generated)

Crate: vox-effort-route · Version: 0.6.0 · Target: Library (lib.rs)

Routes effort-audit findings to verified, drafted enforcement artifacts

Routes effort-audit findings to verified, drafted enforcement artifacts.

See docs/superpowers/specs/2026-05-30-effort-route-design.md.

DependencyKind
vox-actor-runtimeNormal
vox-configNormal
vox-effort-auditNormal
vox-gitNormal
vox-telemetryNormal
ModuleDescription
bucket—
cluster—
config—
embed—
emit—
load—
pipeline—
pricing—
route—
Dependency Topology, diagram 18 nodes, 7 edges: vox-effort-route → vox-actor-runtime; vox-effort-route → vox-config; vox-effort-route → vox-effort-audit; vox-effort-route → vox-git; vox-effort-route → vox-telemetry; vox-cli → vox-effort-route; vox-cli-research → vox-effort-routevox-effort-routevox-actor-runtimevox-configvox-effort-auditvox-gitvox-telemetryvox-clivox-cli-research
Diagram source
graph {
direction: right
node vox_effort_route: "vox-effort-route"
node vox_actor_runtime: "vox-actor-runtime"
vox_effort_route -> vox_actor_runtime
node vox_config: "vox-config"
vox_effort_route -> vox_config
node vox_effort_audit: "vox-effort-audit"
vox_effort_route -> vox_effort_audit
node vox_git: "vox-git"
vox_effort_route -> vox_git
node vox_telemetry: "vox-telemetry"
vox_effort_route -> vox_telemetry
node vox_cli: "vox-cli"
vox_cli -> vox_effort_route
node vox_cli_research: "vox-cli-research"
vox_cli_research -> vox_effort_route
}
Consumer CrateRole
vox-cliDirect workspace dependent
vox-cli-researchDirect workspace dependent
KindSymbolSignatureSourceDescription
structbucket::BucketKeypub struct BucketKeycrates/vox-effort-route/src/bucket.rs#L7-L11—
structbucket::Bucketpub struct Bucketcrates/vox-effort-route/src/bucket.rs#L14-L17—
fnbucket::primary_cratepub fn primary_crate(f: &LoadedFinding) -> Stringcrates/vox-effort-route/src/bucket.rs#L26-L45Derive the owning crate from a finding’s evidence pointers (preferred) or shape histogram (fallback). Returns "<workspace-root>" when no crate path found. Per spec §3 step 2, this is the crate owning the plurality of the finding’s touched paths (evidence pointers), not merely the first one. Ties break deterministically by lexicographically-smallest crate name so the bucket key is stable across runs.
fnbucket::grouppub fn group(findings: Vec<LoadedFinding>) -> Vec<Bucket>crates/vox-effort-route/src/bucket.rs#L61-L75—
traitcluster::Embedderpub trait Embedder: Send + Synccrates/vox-effort-route/src/cluster.rs#L8-L10Abstraction over embedding so tests can assert call counts and determinism.
structcluster::Clusterpub struct Clustercrates/vox-effort-route/src/cluster.rs#L14-L17A sub-cluster of a bucket (or the whole bucket when not split).
fncluster::maybe_splitpub async fn maybe_split(crates/vox-effort-route/src/cluster.rs#L20Split oversized buckets into sub-clusters; pass small buckets through unchanged.
structconfig::EffortRouteConfigpub struct EffortRouteConfigcrates/vox-effort-route/src/config.rs#L7-L22—
structconfig::RouteJudgeConfigpub struct RouteJudgeConfigcrates/vox-effort-route/src/config.rs#L25-L36—
structembed::LlmEmbedderpub struct LlmEmbeddercrates/vox-effort-route/src/embed.rs#L18-L23Embeds finding rationales through the model-agnostic facade. model is resolved upstream by the caller (the orchestrator model registry for the CLI). provider: "auto" defers vendor selection to the facade.
structemit::RecommendationRowpub struct RecommendationRowcrates/vox-effort-route/src/emit/mod.rs#L14-L17One row in recommendations.jsonl.
fnemit::newpub fn new(decision: RemediationDecision) -> Selfcrates/vox-effort-route/src/emit/mod.rs#L20-L25—
enumload::LoadErrorpub enum LoadErrorcrates/vox-effort-route/src/load.rs#L9-L19—
structload::LoadedFindingpub struct LoadedFindingcrates/vox-effort-route/src/load.rs#L23-L25A finding that survived filtering (guaranteed finding.is_some() and score >= threshold).
fnload::readpub fn read(path: &Path, min_waste_score: u8) -> Result<Vec<LoadedFinding>, LoadError>crates/vox-effort-route/src/load.rs#L28-L51Parse, schema-validate, and filter findings.jsonl.
structpipeline::RouteSummarypub struct RouteSummarycrates/vox-effort-route/src/pipeline.rs#L28-L47Roll-up of a single vox audit effort-route run.
fnpipeline::runpub async fn run(crates/vox-effort-route/src/pipeline.rs#L55Run the full effort-route pipeline against an S1 findings.jsonl. Writes recommendations.jsonl, recommendations.md, and verified .proposed artifacts under out_dir. The router and embedder are injected so tests can substitute deterministic mocks and the CLI can wire the facade-backed LLM implementations.
structpricing::ModelRatespub struct ModelRatescrates/vox-effort-route/src/pricing.rs#L9-L13—
fnpricing::cost_usdpub fn cost_usd(&self, prompt_tokens: u64, completion_tokens: u64) -> Option<f64>crates/vox-effort-route/src/pricing.rs#L18-L26Real USD cost for a single judge call, or None when the model is not in the pricing catalog (we never invent a $0.00 for an unpriced model).
enumroute::ArtifactFormpub enum ArtifactFormcrates/vox-effort-route/src/route/mod.rs#L32-L41—
fnroute::staging_extensionpub fn staging_extension(self) -> &'static strcrates/vox-effort-route/src/route/mod.rs#L45-L55Staging-file extension for this form (always ends in .proposed).
fnroute::vox_requiredpub fn vox_required(self) -> boolcrates/vox-effort-route/src/route/mod.rs#L58-L60Whether this form requires the authoring model to be Vox-capable (true only for VoxLang). Such forms are gated out on non-Vox-capable runs.
structroute::DraftedArtifactpub struct DraftedArtifactcrates/vox-effort-route/src/route/mod.rs#L76-L82—
structroute::RemediationDecisionpub struct RemediationDecisioncrates/vox-effort-route/src/route/mod.rs#L85-L109—
fnroute::judge_tokens_usedpub fn judge_tokens_used(&self) -> u64crates/vox-effort-route/src/route/mod.rs#L114-L116Total judge tokens (prompt + completion). Kept as a helper so budget accounting and reporting that key on a single token count still work.
fnroute::budget_skippedpub fn budget_skipped(cluster: &Cluster, cluster_id: &str) -> RemediationDecisioncrates/vox-effort-route/src/route/mod.rs#L121-L149A cluster that was not routed because the judge token budget was already exhausted. Emitted as a None-form, unverified row so the run is honest about what it skipped rather than silently truncating.
structroute::ModelVoxCapabilitypub struct ModelVoxCapability(pub bool)crates/vox-effort-route/src/route/mod.rs#L155Whether the selected judge model can author Vox source. Passed in by the CLI so this crate need not depend on vox-orchestrator’s model registry.
traitroute::Routerpub trait Router: Send + Synccrates/vox-effort-route/src/route/mod.rs#L158-L166—
structroute::MockRouterpub struct MockRoutercrates/vox-effort-route/src/route/mod.rs#L169-L171Deterministic in-memory router for tests.
fnroute::token_sumpub fn token_sum(cost: &vox_effort_audit::hybrid::MeasuredCost) -> u64crates/vox-effort-route/src/route/mod.rs#L237-L251Sum input+output tokens from a MeasuredCost (0 for Unavailable/Ambiguous).
structroute::LlmRouterpub struct LlmRoutercrates/vox-effort-route/src/route/mod.rs#L259-L274Real router wired through vox_actor_runtime::llm::infer_with_retry. All LLM I/O goes through the model-agnostic facade. The model id is resolved upstream (orchestrator model registry) and passed in as resolved_model; no provider hostnames or SDKs leak in here. See AGENTS.md §Model-Agnostic LLM Boundary.