> For clean Markdown of any page, append .md to the page URL. > For a complete documentation index, see https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/llms.txt. > For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/nemo/relay/_mcp/server. # nemo-relay-adaptive > NeMo Relay Adaptive Generated from `cargo doc --no-deps -p nemo-relay -p nemo-relay-adaptive -p nemo-relay-pii-redaction -p nemo-relay-ffi -p nemo-relay-types -p nemo-relay-plugin -p nemo-relay-worker-proto -p nemo-relay-worker`. ## NeMo Relay Adaptive Adaptive config helpers and core-plugin integration for NeMo Relay. Adaptive behavior is enabled through the generic core plugin system. This crate provides the adaptive runtime, persistence abstractions, learner implementations, and Adaptive Cache Governor (ACG) analysis types used to derive and apply runtime hints from observed NeMo Relay executions. ## Re-exports * `pub use config::AcgComponentConfig;` * `pub use config::AdaptiveConfig;` * `pub use config::AdaptiveHintsComponentConfig;` * `pub use config::BackendSpec;` * `pub use config::ResponseCacheConfig;` * `pub use config::StateConfig;` * `pub use config::TelemetryComponentConfig;` * `pub use config::ToolParallelismComponentConfig;` * `pub use context_helpers::LATENCY_SENSITIVITY_POINTER;` * `pub use context_helpers::extract_scope_path;` * `pub use context_helpers::read_manual_latency_sensitivity;` * `pub use context_helpers::resolve_agent_id;` * `pub use context_helpers::resolve_shared_parent_scope_identity;` * `pub use context_helpers::set_latency_sensitivity;` * `pub use error::AdaptiveError;` * `pub use error::Result;` * `pub use redis::RedisBackend;` * `pub use response_cache::config::ToolCacheConfig;` * `pub use response_cache::config::ToolClass;` * `pub use response_cache::config::ToolOverride;` * `pub use response_cache::RESPONSE_CACHE_MARK;` * `pub use response_cache::ResponseCacheKeyStrategy;` * `pub use storage::erased::AnyBackend;` * `pub use storage::memory::InMemoryBackend;` * `pub use storage::traits::StorageBackend;` * `pub use storage::traits::StorageBackendDyn;` ## Modules * [acg](/reference/api/rust-library-reference/nemo-relay-adaptive/acg): Canonical Adaptive Cache Governor (ACG) module surface exposed from the adaptive crate. * [acg\_component](/reference/api/rust-library-reference/nemo-relay-adaptive/acg_component): Adaptive Cache Governor (ACG) request and execution intercept helpers for the adaptive runtime. * [acg\_learner](/reference/api/rust-library-reference/nemo-relay-adaptive/acg_learner): Adaptive Cache Governor (ACG) learner for the adaptive telemetry pipeline. * [acg\_profile](/reference/api/rust-library-reference/nemo-relay-adaptive/acg_profile): Derives stable Adaptive Cache Governor (ACG) profile keys from structured LLM requests. * [adaptive\_hints\_intercept](/reference/api/rust-library-reference/nemo-relay-adaptive/adaptive_hints_intercept): AdaptiveHintsIntercept: opt-in LLM request intercept that injects AgentHints from HotCache trie. * [cache\_diagnostics](/reference/api/rust-library-reference/nemo-relay-adaptive/cache_diagnostics): Runtime-local cache miss request facts and diagnostics tracking. * [config](/reference/api/rust-library-reference/nemo-relay-adaptive/config): Canonical adaptive config and diagnostics types. * [context\_helpers](/reference/api/rust-library-reference/nemo-relay-adaptive/context_helpers): Context helpers for reading scope metadata on the intercept hot path. * [drain](/reference/api/rust-library-reference/nemo-relay-adaptive/drain): Background drain task for async telemetry processing. * [error](/reference/api/rust-library-reference/nemo-relay-adaptive/error): Error types for the nemo-relay-adaptive crate. * [intercepts](/reference/api/rust-library-reference/nemo-relay-adaptive/intercepts): Intercept factories for the `nemo-relay-adaptive` crate, including Adaptive Cache Governor (ACG) intercepts. * [learner](/reference/api/rust-library-reference/nemo-relay-adaptive/learner): Learning primitives and built-in learner implementations. Adaptive learners that derive runtime hints from observed executions. * [plugin\_component](/reference/api/rust-library-reference/nemo-relay-adaptive/plugin_component): Core plugin integration for the adaptive runtime. * [redis](/reference/api/rust-library-reference/nemo-relay-adaptive/redis): Redis-backed [`StorageBackend`](/reference/api/rust-library-reference/nemo-relay-adaptive/storage/traits/trait-storagebackend) implementation. * [response\_cache](/reference/api/rust-library-reference/nemo-relay-adaptive/response_cache): Opt-in exact-match LLM response and tool-result cache. Opt-in cache for LLM responses and tool results: a feature of the adaptive plugin, configured through [`crate::config::AdaptiveConfig::response_cache`](/reference/api/rust-library-reference/nemo-relay-adaptive/config/struct-adaptiveconfig). * [storage](/reference/api/rust-library-reference/nemo-relay-adaptive/storage): Storage backends and backend traits for adaptive state persistence. Storage backends for adaptive runtime state and learned artifacts. * [subscriber](/reference/api/rust-library-reference/nemo-relay-adaptive/subscriber): Event subscriber factory and event-to-record mapping helpers. * [tool\_parallelism\_learner](/reference/api/rust-library-reference/nemo-relay-adaptive/tool_parallelism_learner): Learner that derives tool fan-out plans from observed runs. Learner that derives tool parallelism plans from observed runs. * [trie](/reference/api/rust-library-reference/nemo-relay-adaptive/trie): Prediction trie data structures for the online learning engine. * [types](/reference/api/rust-library-reference/nemo-relay-adaptive/types): Serializable adaptive data models shared across runtime components. Serializable adaptive runtime data models. ## Structs * [AdaptiveRuntime](/reference/api/rust-library-reference/nemo-relay-adaptive/struct-adaptiveruntime): Hosted adaptive runtime that registers NeMo Relay plugin components. > NeMo Relay Adaptive ## Docs - [Module acg](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg.md): Canonical Adaptive Cache Governor (ACG) module surface exposed from the adaptive crate. - [Module anthropic_plugin](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/anthropic_plugin.md): Anthropic cache plugin for the Adaptive Cache Governor (ACG) system. - [Struct Anthropic Cache Plugin](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/anthropic_plugin/struct-anthropiccacheplugin.md): Anthropic-specific provider plugin for cache_control breakpoint injection. - [Module canonicalize](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/canonicalize.md): RFC 8785 JSON Canonicalization and whitespace normalization for cache stability. - [Function canonicalize_json](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/canonicalize/fn-canonicalize-json.md): Canonicalize a JSON string per RFC 8785 (JCS). - [Function canonicalize_value](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/canonicalize/fn-canonicalize-value.md): Canonicalize a `serde_json::Value` per RFC 8785. - [Function normalize_whitespace](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/canonicalize/fn-normalize-whitespace.md): Normalize whitespace in text content. - [Function sha256_hex](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/canonicalize/fn-sha256-hex.md): Compute SHA-256 hex digest with `"sha256:"` prefix. - [Module capability](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/capability.md): Capability registry tracking per-backend and per-model-family supported features. - [Struct Backend Capabilities](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/capability/struct-backendcapabilities.md): Capabilities of a specific backend provider. - [Struct Cache Economics](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/capability/struct-cacheeconomics.md): Provider/model-specific cache economics used by the internal planner. - [Struct Capability Registry](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/capability/struct-capabilityregistry.md): Registry holding capabilities for all known backends. - [Struct Model Family Capabilities](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/capability/struct-modelfamilycapabilities.md): Per-model-family capability overrides within a backend. - [Enum Provider Feature](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/capability/enum-providerfeature.md): Feature that a backend or model family may support. - [Module ir_builder](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/ir_builder.md): AnnotatedLlmRequest to PromptIR construction pipeline. - [Function build_prompt_ir](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/ir_builder/fn-build-prompt-ir.md): Build a normalized `PromptIR` from an annotated LLM request. - [Module openai_plugin](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/openai_plugin.md): OpenAI cache plugin for the Adaptive Cache Governor (ACG) system. - [Struct OpenAI Cache Plugin](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/openai_plugin/struct-openaicacheplugin.md): OpenAI-specific provider plugin for deterministic JSON serialization. - [Module passthrough](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/passthrough.md): Passthrough plugin that applies no transformations. - [Struct Passthrough Plugin](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/passthrough/struct-passthroughplugin.md): A no-op provider plugin that passes requests through unchanged. - [Module plugin](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/plugin.md): Provider plugin trait and input/output types for the Adaptive Cache Governor (ACG) system. - [Struct Plugin Input](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/plugin/struct-plugininput.md): Input data provided to a plugin for translation. - [Struct Plugin Output](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/plugin/struct-pluginoutput.md): Output produced by a plugin after translation. - [Trait Provider Plugin](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/plugin/trait-providerplugin.md): Provider plugin trait for backend-specific translation. - [Module plugin_registry](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/plugin_registry.md): Plugin registry for storing and retrieving provider plugins. - [Struct Plugin Registry](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/plugin_registry/struct-pluginregistry.md): Registry for storing and retrieving provider plugins. - [Module policy](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/policy.md): Policy types for the Adaptive Cache Governor (ACG) system. - [Struct Archetype Routing Override](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/policy/struct-archetyperoutingoverride.md): Routing override for one observed session archetype. - [Struct Cache Policy](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/policy/struct-cachepolicy.md): Policy controlling when ACG outputs can be cached and reused. - [Struct Policy Envelope](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/policy/struct-policyenvelope.md): Versioned wrapper for an ACG policy document. - [Struct Retention Policy](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/policy/struct-retentionpolicy.md): Policy controlling how long learned artifacts should be retained. - [Struct Rewrite Policy](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/policy/struct-rewritepolicy.md): Policy controlling what prompt rewrites the runtime can apply automatically. - [Struct Routing Policy](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/policy/struct-routingpolicy.md): Policy controlling model-class routing decisions. - [Struct Scope Retention Override](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/policy/struct-scoperetentionoverride.md): Retention-tier override for one sharing scope. - [Enum Transformation Class](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/policy/enum-transformationclass.md): Rewrite category that an ACG policy can allow or deny. - [Module profile](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/profile.md): Behavioral profile types for the Adaptive Cache Governor (ACG) system. - [Struct Behavioral Profile](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/profile/struct-behavioralprofile.md): Aggregated behavioral profile derived from observed runs. - [Struct Block Stability Score](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/profile/struct-blockstabilityscore.md): Stability analysis result for one prompt block. - [Struct Distribution Summary](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/profile/struct-distributionsummary.md): Percentile summary for a duration-like metric. - [Struct Parallelism Pattern](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/profile/struct-parallelismpattern.md): Summary of observed parallelism behavior across runs. - [Struct Tool Usage Phase](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/profile/struct-toolusagephase.md): One phase of tool usage observed in an agent workflow. - [Enum Session Archetype](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/profile/enum-sessionarchetype.md): Coarse behavioral archetype inferred from observed runs. - [Enum Stability Class](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/profile/enum-stabilityclass.md): Stability bucket assigned to a prompt block. - [Module prompt_ir](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/prompt_ir.md): Prompt Intermediate Representation (IR) types for the Adaptive Cache Governor (ACG) system. - [Struct Prompt Block](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/prompt_ir/struct-promptblock.md): A single addressable block within the Prompt IR. - [Struct PromptIR](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/prompt_ir/struct-promptir.md): Prompt Intermediate Representation - the full decomposed prompt. - [Struct SpanId](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/prompt_ir/struct-spanid.md): Stable span identifier for addressable prompt blocks. - [Struct Tokenization Metadata](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/prompt_ir/struct-tokenizationmetadata.md): Token count metadata for a prompt block. - [Struct Tool Schema Hash](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/prompt_ir/struct-toolschemahash.md): Hash fingerprint of a tool schema definition. - [Enum Block Content Type](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/prompt_ir/enum-blockcontenttype.md): Content type discriminant for a prompt block. - [Enum Prompt Role](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/prompt_ir/enum-promptrole.md): Role of a prompt block within the conversation. - [Enum Provenance Label](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/prompt_ir/enum-provenancelabel.md): Origin label for a prompt block. - [Enum Sensitivity Label](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/prompt_ir/enum-sensitivitylabel.md): Sensitivity classification for a prompt block. - [Module retention](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/retention.md): Retention intent generation from observed session timing distributions. - [Struct Retention Thresholds](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/retention/struct-retentionthresholds.md): Thresholds used to map observed timing into a retention tier. - [Function generate_retention_intent](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/retention/fn-generate-retention-intent.md): Generate a retention intent from observed session timing distributions. - [Function generate_retention_intent_default](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/retention/fn-generate-retention-intent-default.md): Generate a retention intent with the default thresholds. - [Module stability](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/stability.md): Stability analysis for prompt blocks across multiple observations. - [Struct Stability Analysis Result](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/stability/struct-stabilityanalysisresult.md): Result of analyzing prompt stability across a set of observations. - [Struct Stability Thresholds](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/stability/struct-stabilitythresholds.md): Thresholds controlling prompt-block stability classification. - [Function analyze_stability](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/stability/fn-analyze-stability.md): Analyze prompt-block stability across multiple observations. - [Module telemetry](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/telemetry.md): Cache performance telemetry types for the Adaptive Cache Governor (ACG) system. - [Struct Cache HitRate](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/telemetry/struct-cachehitrate.md): Aggregated cache hit rate over a time window. - [Struct Cache Miss Diagnosis](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/telemetry/struct-cachemissdiagnosis.md): Structured diagnosis for a cache miss. - [Struct Cache Request Facts](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/telemetry/struct-cacherequestfacts.md): Request-time facts used to classify a cache miss without leaking prompt text. - [Struct Cache Telemetry Event](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/telemetry/struct-cachetelemetryevent.md): Per-call cache telemetry event. - [Enum Cache Miss Evidence](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/telemetry/enum-cachemissevidence.md): Typed evidence for a cache miss diagnosis. - [Enum Cache Miss Reason](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/telemetry/enum-cachemissreason.md): Reason why a cache miss occurred. - [Enum Cache Telemetry Provider](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/telemetry/enum-cachetelemetryprovider.md): Cache telemetry provider identity for canonical `Usage` normalization. - [Module types](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types.md): Core data types for the Adaptive Cache Governor (ACG) crate. - [Struct Agent Identity](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/struct-agentidentity.md): Identity model for an agent type. - [Struct Cache Stability Intent](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/struct-cachestabilityintent.md): Cache stability analysis results for a prompt region. - [Struct Compression Intent](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/struct-compressionintent.md): Content compression intent. - [Struct Content Extraction Intent](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/struct-contentextractionintent.md): Content extraction intent for variable content detection. - [Struct Intent Outcome](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/struct-intentoutcome.md): Records the outcome of translating a single optimization intent. - [Struct Model Routing Intent](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/struct-modelroutingintent.md): Model routing and selection intent. - [Struct Optimization Intent Bundle](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/struct-optimizationintentbundle.md): A bundle of optimization intents for a single request. - [Struct Placement Intent](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/struct-placementintent.md): Context placement optimization intent. - [Struct Priority Intent](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/struct-priorityintent.md): Latency and priority routing intent. - [Struct Retention Intent](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/struct-retentionintent.md): Cache retention policy intent. - [Struct Serialization Intent](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/struct-serializationintent.md): Serialization and fanout optimization intent. - [Struct Tool Scope Intent](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/struct-toolscopeintent.md): Tool scope and phase management intent. - [Struct Translation Report](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/struct-translationreport.md): A plugin's complete report on how it handled an intent bundle. - [Enum Intent Type](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/enum-intenttype.md): Discriminant enum for intent types (used in translation report outcomes). - [Enum Model Class](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/enum-modelclass.md): Model complexity/criticality class for routing. - [Enum Optimization Intent](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/enum-optimizationintent.md): A single optimization intent emitted by a behavioral model. - [Enum Placement Target](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/enum-placementtarget.md): Target location for context placement. - [Enum Reason Code](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/enum-reasoncode.md): Machine-readable reason for the translation outcome. - [Enum Retention Tier](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/enum-retentiontier.md): Retention tier for cached state. - [Enum Sharing Scope](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/enum-sharingscope.md): Sharing scope for cached content - stability does not imply shareability. Default is `Session` per security requirements. - [Enum Translation Status](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/types/enum-translationstatus.md): Outcome status for a single intent translation. - [Module variable_extractor](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/variable_extractor.md): Variable content detection and extraction from prompt blocks. - [Struct Extracted Variable](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/variable_extractor/struct-extractedvariable.md): One extracted variable occurrence within a prompt block. - [Struct Extraction Result](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/variable_extractor/struct-extractionresult.md): Variable extraction result for one prompt block. - [Struct Variable Pattern](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/variable_extractor/struct-variablepattern.md): Regex-based pattern used to detect variable content. - [Enum Variable Category](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/variable_extractor/enum-variablecategory.md): Category assigned to an extracted variable. - [Function default_variable_patterns](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/variable_extractor/fn-default-variable-patterns.md): Return the default regex patterns used for variable extraction. - [Function extract_variables](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/variable_extractor/fn-extract-variables.md): Extract variables from one content string. - [Function extract_variables_from_blocks](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/variable_extractor/fn-extract-variables-from-blocks.md): Extract variables from a set of prompt blocks. - [Module error](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/error.md): Error types for the Adaptive Cache Governor (ACG) crate. - [Enum AcgError](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/error/enum-acgerror.md): The error type for all Adaptive Cache Governor (ACG) operations. - [Type Alias Result](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/error/type-result.md): A specialized `Result` type for ACG operations. - [Constant MIN_ACG_OBSERVATIONS](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg/constant-min-acg-observations.md): Minimum observations required before Adaptive Cache Governor (ACG) emits optimization intents. - [Module acg_component](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg_component.md): Adaptive Cache Governor (ACG) request and execution intercept helpers for the adaptive runtime. - [Module acg_learner](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg_learner.md): Adaptive Cache Governor (ACG) learner for the adaptive telemetry pipeline. - [Struct AcgLearner](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg_learner/struct-acglearner.md): Learner that derives prompt stability state for ACG. - [Module acg_profile](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/acg_profile.md): Derives stable Adaptive Cache Governor (ACG) profile keys from structured LLM requests. - [Module adaptive_hints_intercept](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/adaptive_hints_intercept.md): AdaptiveHintsIntercept: opt-in LLM request intercept that injects AgentHints from HotCache trie. - [Struct Adaptive Hints Intercept](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/adaptive_hints_intercept/struct-adaptivehintsintercept.md): Opt-in LLM request intercept that injects `AgentHints` into request headers from the prediction trie in `HotCache`. - [Module cache_diagnostics](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/cache_diagnostics.md): Runtime-local cache miss request facts and diagnostics tracking. - [Struct Cache Diagnostics Tracker](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/cache_diagnostics/struct-cachediagnosticstracker.md): Runtime-local miss diagnosis tracker. - [Function build_cache_request_facts](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/cache_diagnostics/fn-build-cache-request-facts.md): Builds canonical request facts for cache miss diagnosis from the live runtime state. - [Module config](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/config.md): Canonical adaptive config and diagnostics types. - [Struct AcgComponent Config](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/config/struct-acgcomponentconfig.md): Typed helper for the built-in Adaptive Cache Governor (ACG) component. - [Struct Response Cache Config](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/config/struct-responsecacheconfig.md): Configuration for the adaptive plugin's LLM response and opt-in tool-result cache feature. - [Struct Adaptive Config](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/config/struct-adaptiveconfig.md): Canonical config document for the adaptive plugin component. - [Struct Adaptive Hints Component Config](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/config/struct-adaptivehintscomponentconfig.md): Typed helper for adaptive hints settings. - [Struct Backend Spec](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/config/struct-backendspec.md): Dynamic backend selection. `config` is backend-specific. - [Struct State Config](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/config/struct-stateconfig.md): Shared state configuration consumed by adaptive features that need persistence. - [Struct Telemetry Component Config](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/config/struct-telemetrycomponentconfig.md): Typed helper for telemetry settings. - [Struct Tool Parallelism Component Config](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/config/struct-toolparallelismcomponentconfig.md): Typed helper for tool parallelism settings. - [Module context_helpers](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/context_helpers.md): Context helpers for reading scope metadata on the intercept hot path. - [Struct Shared Parent Scope Identity](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/context_helpers/struct-sharedparentscopeidentity.md): Session-local scope identity used to coordinate warm-first cohorts. - [Constant LATENCY_SENSITIVITY_POINTER](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/context_helpers/constant-latency-sensitivity-pointer.md): Metadata key path for manual latency sensitivity annotation. - [Function extract_scope_path](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/context_helpers/fn-extract-scope-path.md): Extracts the current function call path from the NeMo Relay scope stack. - [Function read_manual_latency_sensitivity](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/context_helpers/fn-read-manual-latency-sensitivity.md): Reads the maximum manual latency sensitivity from all scopes in the current scope stack. - [Function resolve_agent_id](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/context_helpers/fn-resolve-agent-id.md): Resolves the agent ID from the current scope stack. - [Function resolve_shared_parent_scope_identity](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/context_helpers/fn-resolve-shared-parent-scope-identity.md): Resolves the session-local identity used by warm-first cohort coordination. - [Function set_latency_sensitivity](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/context_helpers/fn-set-latency-sensitivity.md): Sets latency sensitivity on the current (top) scope using max-merge semantics. - [Module drain](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/drain.md): Background drain task for async telemetry processing. - [Module error](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/error.md): Error types for the nemo-relay-adaptive crate. - [Enum Adaptive Error](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/error/enum-adaptiveerror.md): The error type for all nemo-relay-adaptive operations. - [Type Alias Result](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/error/type-result.md): A specialized `Result` type for nemo-relay-adaptive operations. - [Module intercepts](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/intercepts.md): Intercept factories for the `nemo-relay-adaptive` crate, including Adaptive Cache Governor (ACG) intercepts. - [Constant AGENT_HINTS_HEADER_KEY](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/intercepts/constant-agent-hints-header-key.md): Header key used to propagate serialized adaptive agent hints. - [Module learner](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/learner.md): Learning primitives and built-in learner implementations. Adaptive learners that derive runtime hints from observed executions. - [Module latency](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/learner/latency.md): Learner that builds latency sensitivity hints from run history. Latency-sensitivity learner implementation. - [Struct Latency Sensitivity Learner](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/learner/latency/struct-latencysensitivitylearner.md): Learner that derives default latency sensitivity hints from run history. - [Function compute_default_hints](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/learner/latency/fn-compute-default-hints.md): Compute default agent hints from the root trie prediction. - [Module traits](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/learner/traits.md): Common learner trait implemented by adaptive background processors. Shared learner traits for adaptive background processing. - [Trait Learner](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/learner/traits/trait-learner.md): Background learner that updates adaptive state from observed runs. - [Module plugin_component](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/plugin_component.md): Core plugin integration for the adaptive runtime. - [Struct Component Spec](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/plugin_component/struct-componentspec.md): One configured adaptive component. - [Constant ADAPTIVE_PLUGIN_KIND](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/plugin_component/constant-adaptive-plugin-kind.md): The plugin kind registered by the adaptive crate. - [Function deregister_adaptive_component](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/plugin_component/fn-deregister-adaptive-component.md): Deregisters the adaptive component kind from the core plugin registry. - [Function register_adaptive_component](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/plugin_component/fn-register-adaptive-component.md): Registers the adaptive component kind in the core plugin registry. - [Module redis](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/redis.md): Redis-backed `StorageBackend` implementation. - [Struct Redis Backend](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/redis/struct-redisbackend.md): A Redis-backed storage backend for cross-process shared state. - [Module response_cache](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/response_cache.md): Opt-in exact-match LLM response and tool-result cache. Opt-in cache for LLM responses and tool results: a feature of the adaptive plugin, configured through `crate::config::AdaptiveConfig::response_cache`. - [Module config](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/response_cache/config.md): Backend selection for the adaptive plugin's `response_cache` feature. - [Struct Backend Config](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/response_cache/config/struct-backendconfig.md): Backend selection mirroring the adaptive `crate::config::BackendSpec` shape. - [Enum Response Cache KeyStrategy](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/response_cache/config/enum-responsecachekeystrategy.md): Strategy for deriving an LLM response-cache key. - [Struct Tool Class](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/response_cache/config/struct-toolclass.md): Policy shared by a class of tools. - [Struct Tool Cache Config](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/response_cache/config/struct-toolcacheconfig.md): Opt-in tool-result cache configuration. - [Struct Tool Override](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/response_cache/config/struct-tooloverride.md): Per-tool refinement applied after class resolution. - [Constant DEFAULT_MAX_BYTES](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/response_cache/config/constant-default-max-bytes.md): Default in-memory byte budget: 256 MiB. - [Constant RESPONSE_CACHE_MARK](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/response_cache/constant-response-cache-mark.md): Mark-event name emitted on every cache decision. - [Module storage](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/storage.md): Storage backends and backend traits for adaptive state persistence. Storage backends for adaptive runtime state and learned artifacts. - [Module erased](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/storage/erased.md): Erased backend alias used by runtime configuration and dependency injection. Type-erased storage backend aliases. - [Type Alias AnyBackend](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/storage/erased/type-anybackend.md): Type-erased adaptive storage backend. - [Module memory](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/storage/memory.md): In-memory backend useful for tests and local-only runtime state. In-memory adaptive storage backend. - [Struct InMemory Backend](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/storage/memory/struct-inmemorybackend.md): In-memory backend storing adaptive state in process-local hash maps. - [Module traits](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/storage/traits.md): Async storage traits implemented by adaptive persistence backends. Storage traits used by adaptive backends. - [Trait Storage Backend](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/storage/traits/trait-storagebackend.md): Minimal async storage interface required by the adaptive runtime. - [Trait Storage Backend Dyn](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/storage/traits/trait-storagebackenddyn.md): Object-safe storage interface used by the adaptive runtime. - [Module subscriber](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/subscriber.md): Event subscriber factory and event-to-record mapping helpers. - [Module tool_parallelism_learner](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/tool_parallelism_learner.md): Learner that derives tool fan-out plans from observed runs. Learner that derives tool parallelism plans from observed runs. - [Struct Tool Parallelism Learner](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/tool_parallelism_learner/struct-toolparallelismlearner.md): Learner that discovers tool fan-out groups from run telemetry. - [Module trie](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/trie.md): Prediction trie data structures for the online learning engine. - [Module accumulator](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/trie/accumulator.md): Streaming statistics accumulator using Welford's algorithm and TDigest. - [Struct Accumulator State](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/trie/accumulator/struct-accumulatorstate.md): Maps trie path strings to their node accumulators. - [Struct Node Accumulators](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/trie/accumulator/struct-nodeaccumulators.md): Per-node accumulators for all metric types, keyed by call index. - [Struct Running Stats](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/trie/accumulator/struct-runningstats.md): Streaming statistics tracker combining Welford's online algorithm for mean/variance with a TDigest for streaming percentile estimation. - [Module builder](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/trie/builder.md): Prediction trie builder with incremental accumulator merge. - [Struct Prediction Trie Builder](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/trie/builder/struct-predictiontriebuilder.md): Builds a `PredictionTrieNode` tree from `RunRecord`s via incremental accumulator merge. - [Struct Sensitivity Config](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/trie/builder/struct-sensitivityconfig.md): Configuration for auto-sensitivity scoring. - [Module data_models](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/trie/data_models.md): Core prediction trie data types with NAT wire-format compatible serialization. - [Struct LlmCall Prediction](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/trie/data_models/struct-llmcallprediction.md): Predictions for an LLM call at a given position in the call hierarchy. - [Struct Prediction Metrics](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/trie/data_models/struct-predictionmetrics.md): Aggregated statistics for a single metric from profiler data. - [Struct Prediction Trie Node](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/trie/data_models/struct-predictiontrienode.md): A node in the prediction trie representing a function in the call hierarchy. - [Module lookup](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/trie/lookup.md): Prediction trie lookup with three-level fallback chain. - [Struct Prediction Trie Lookup](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/trie/lookup/struct-predictiontrielookup.md): Looks up predictions in a prediction trie with graceful fallback. - [Module serialization](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/trie/serialization.md): NAT wire-format serialization for prediction tries. - [Struct Trie Envelope](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/trie/serialization/struct-trieenvelope.md): Versioned envelope wrapping a prediction trie for JSON persistence. - [Constant CURRENT_VERSION](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/trie/serialization/constant-current-version.md): Version string for the trie wire format. - [Module types](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/types.md): Serializable adaptive data models shared across runtime components. Serializable adaptive runtime data models. - [Module cache](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/types/cache.md): Hot-cache structures used by the adaptive runtime on the intercept path. Hot-cache state shared by adaptive runtime features. - [Struct HotCache](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/types/cache/struct-hotcache.md): In-memory cache of adaptive artifacts needed on the hot path. - [Module metadata](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/types/metadata.md): Metadata and hint payloads attached to adaptive execution plans. Metadata and hint payload types used by adaptive planning. - [Struct Agent Hints](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/types/metadata/struct-agenthints.md): Runtime hint bundle exposed to downstream integrations. - [Struct Metadata Envelope](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/types/metadata/struct-metadataenvelope.md): Metadata template attached to an adaptive execution plan. - [Struct Parallel Hint](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/types/metadata/struct-parallelhint.md): Hint describing one tool's membership in a parallel-execution cohort. - [Type Alias Json](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/types/metadata/type-json.md): JSON value alias used by adaptive metadata payloads. - [Module plan](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/types/plan.md): Execution-plan types describing discovered tool parallelism. Adaptive execution-plan data models. - [Struct Execution Plan](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/types/plan/struct-executionplan.md): Learned execution plan for an agent. - [Struct Parallel Group](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/types/plan/struct-parallelgroup.md): Group of tools that have been observed to run in parallel. - [Module records](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/types/records.md): Run and call record types collected by the telemetry pipeline. Run and call record types collected by the adaptive telemetry pipeline. - [Struct Call Record](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/types/records/struct-callrecord.md): Telemetry record for a single tool or LLM call. - [Struct RunRecord](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/types/records/struct-runrecord.md): Telemetry record for one observed agent run. - [Enum Call Kind](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/types/records/enum-callkind.md): Kind of runtime call captured in adaptive telemetry. - [Struct Adaptive Runtime](https://docs.nvidia.com/nemo/relay/v0.10.0/reference/api/rust-library-reference/nemo-relay-adaptive/struct-adaptiveruntime.md): Hosted adaptive runtime that registers NeMo Relay plugin components.