Crate everruns_provider 

Lean provider and LLM abstractions shared by the Everruns Framework and provider crates.

This crate owns credential-free model identity, the ChatDriver boundary, provider assembly, shared protocol drivers, stream/retry helpers, typed IDs, credential schemas, and the LLM error taxonomy. It is a focused implementation crate in the Everruns ecosystem. Low-level consumers import provider-owned contracts from this crate directly; Framework users normally use the everruns facade.

Model profile types and data (ModelProfile, ServiceKind, and the built-in profile registry) live in everruns-model-profiles and are re-exported here for source compatibility.

Example

use everruns_provider::ModelSpec;

let model = ModelSpec::on("company-gateway", "assistant-v2");
assert_eq!(model.provider.as_str(), "company-gateway");
assert_eq!(model.model, "assistant-v2");

Re-exports

pub use compact::CompactContent;

pub use compact::CompactContentPart;

pub use compact::CompactInputItem;

pub use compact::CompactOutputItem;

pub use compact::CompactRequest;

pub use compact::CompactResponse;

pub use compact::CompactUsage;

pub use compact::messages_to_compact_input;

pub use credential_provider::CredentialProvider;

pub use credential_provider::EnvCredentialProvider;

pub use credential_provider::ProviderCredentials;

pub use credential_schema::CredentialFormSchema;

pub use credential_schema::assemble_credential_document;

pub use credential_schema::parse_credential_document;

pub use driver_registry::BoxedChatDriver;

pub use driver_registry::BoxedEmbeddingsDriver;

pub use driver_registry::ChatDriver;

pub use driver_registry::DiscoveredModel;

pub use driver_registry::DriverDescriptor;

pub use driver_registry::DriverFactory;

pub use driver_registry::DriverId;

pub use driver_registry::DriverOAuthConfig;

pub use driver_registry::DriverOAuthFlow;

pub use driver_registry::DriverRegistry;

pub use driver_registry::EmbedRequest;

pub use driver_registry::EmbedResponse;

pub use driver_registry::EmbeddingsDriver;

pub use driver_registry::EmbeddingsDriverError;

pub use driver_registry::EmbeddingsDriverFactory;

pub use driver_registry::LlmCallConfig;

pub use driver_registry::LlmCallConfigBuilder;

pub use driver_registry::LlmCompletionMetadata;

pub use driver_registry::LlmContentPart;

pub use driver_registry::LlmMessage;

pub use driver_registry::LlmMessageContent;

pub use driver_registry::LlmMessageRole;

pub use driver_registry::LlmResponse;

pub use driver_registry::LlmResponseStream;

pub use driver_registry::LlmStreamError;

pub use driver_registry::LlmStreamEvent;

pub use driver_registry::ProviderConfig;

pub use driver_registry::ProviderMetadata;

pub use driver_registry::ProviderOpaqueContext;

pub use driver_registry::fold_system_messages;

pub use error::AgentLoopError;

pub use error::BillingPressureReason;

pub use error::FileSystemError;

pub use error::FileSystemErrorClass;

pub use error::LlmError;

pub use error::LlmErrorKind;

pub use error::Result;

pub use error::StoreResultExt;

pub use error::classify_fs_error;

pub use error::from_json;

pub use error::json_val;

pub use execution_phase::ExecutionPhase;

pub use execution_phase::PhaseSource;

pub use llm_retry::LlmRetryConfig;

pub use llm_retry::RateLimitInfo;

pub use llm_retry::RateLimitType;

pub use llm_retry::RetryMetadata;

pub use model::Model;

pub use model::ModelSource;

pub use model::ModelWithProvider;

pub use model_discovery::list_openai_compatible_models;

pub use model_discovery::DiscoveredProviderModel;

pub use model_discovery::ModelSearchMatch;

pub use model_discovery::ModelSearchResult;

pub use model_discovery::RankedDiscoveredModels;

pub use model_discovery::discover_provider_models;

pub use model_discovery::enrich_with_profiles;

pub use model_discovery::match_models;

pub use model_discovery::normalize_and_enrich;

pub use model_discovery::rank_discovered_models;

pub use model_discovery::search_provider_models;

pub use model_profiles::get_model_profile;

pub use model_profiles::get_model_vendor;

pub use model_spec::ModelSpec;

pub use model_spec::UnknownProvider;

pub use openai_protocol::OpenAIProtocolChatDriver;

pub use openresponses_protocol::OpenResponsesProtocolChatDriver;

pub use openresponses_protocol::OpenResponsesRequestExtension;

pub use provider::Provider as ProviderRecord;

pub use provider::ProviderStatus;

pub use provider::ProviderTraceConfig;

pub use reasoning::ReasoningContentPart;

pub use reasoning::ReasoningText;

pub use runtime_provider::BearerAuth;

pub use runtime_provider::Provider;

pub use runtime_provider::ProviderAuth;

pub use runtime_provider::ProviderAuthRequest;

pub use runtime_provider::ProviderEndpoint;

pub use runtime_provider::ProviderKey;

pub use runtime_provider::ProviderRegistry;

pub use runtime_provider::ResolvedProviderRequest;

pub use runtime_provider::RuntimeProvider;

pub use runtime_provider::RuntimeProviderRegistry;

pub use runtime_provider::StaticHeaderAuth;

pub use tool_types::BuiltinTool;

pub use tool_types::CONFIRM_URL_ELICITATION_TOOL;

pub use tool_types::ClientSideTool;

pub use tool_types::DeferrablePolicy;

pub use tool_types::SideEffectClass;

pub use tool_types::ToolCall;

pub use tool_types::ToolDefinition;

pub use tool_types::ToolHints;

pub use tool_types::ToolPolicy;

pub use tool_types::ToolResult;

pub use tool_types::ToolResultImage;

pub use tool_types::URL_ELICITATION_REQUIRED_CODE;

pub use tool_types::UrlElicitationRequired;

pub use url_validation::UrlValidationError;

pub use url_validation::is_blocked_ip;

pub use url_validation::validate_safe_url;

pub use url_validation::validate_url_dns_pinned;

pub use user_facing_error::ErrorDisclosure;

pub use user_facing_error::UserFacingError;

pub use user_facing_error::UserFacingErrorContext;

pub use user_facing_error::UserFacingErrorFields;

pub use user_facing_error::classify_runtime_error_message;

pub use user_facing_error::codes as user_facing_error_codes;

pub use user_facing_error::is_provider_quota_message;

pub use user_facing_error::is_usage_limit_message;

pub use user_facing_error::metadata_keys as user_facing_error_metadata_keys;

pub use user_facing_error::parse_usage_limit_reset_at;

pub use user_facing_error::trim_error_chain_prefixes;

Modules

compact

Transport-neutral native compaction request and response contracts.

credential_provider

credential_schema

driver_helpers

driver_registry

error

execution_phase

Execution phase for assistant messages in multi-step tool-calling flows.

llm_retry

model

model_discovery

Provider model discovery and display ranking.

model_profiles

model_spec

Credential-free model selection.

native_async

Native asynchronous calls are pending provider exchanges, not background-run handles.

openai_compat

OpenAI model constraints checked before a request reaches the network.

openai_protocol

openresponses_protocol

openresponses_types

provider

reasoning

Provider reasoning artifacts.

reasoning_updates

Conversation-scoped Astra configuration, distinct from the request baseline.

runtime_provider

Runtime providers: service identity, endpoint, authentication, and wire driver.

stream_accumulator

stream_reconnect

tool_schema_compat

Provider-specific JSON Schema compatibility at the LLM request boundary.

tool_types

typed_id

url_validation

user_facing_error

Structs

CostTier

A pricing tier that activates above a context token threshold. For example, OpenAI charges higher rates for prompts exceeding 200K tokens.

ModelCost

Cost information for the model (per million tokens)

ModelLimits

Token limits for the model

ModelModalities

Model modalities for input and output

ModelProfile

LLM Model Profile describing model capabilities Based on models.dev structure (https://models.dev/api.json)

ReasoningEffortConfig

Reasoning effort configuration for a model

ReasoningEffortValue

Named reasoning effort value for UI display

Enums

Modality

Modality type (text, image, audio, video)

ModelVendor

Vendor / brand that authored a model. Independent of the provider type that serves it (the same model may be offered by several providers or gateways). Primarily drives UI iconography.

ReasoningEffort

Reasoning effort level for models that support it

ServiceKind

A typed service a provider driver can offer (see knowledge/foundations/providers.md).

Functions

install_default_crypto_provider

Select the workspace crypto backend as the process-wide rustls provider.