inference

Oct 27, 2026

Changes in this version

+ var ErrReuseUncertain = errors.New("reuse verification outcome is uncertain; inspect its receipt before retrying")

Oct 27, 2026

Changes in this version

type OllamaProvider

+ func (p *OllamaProvider) Close()

type Result

+ Resolution *ReuseResolution

+ ReuseKind string

+ type ReuseResolution struct

+ PolicyID string

+ Response Response

+ Source ReuseSource

+ VerificationKey string

+ type ReuseResolver interface

+ Identity func() string

+ Resolve func(context.Context, Spec) (*ReuseResolution, error)

+ type ReuseSource struct

+ Key string

+ OutputSHA256 string

+ RequestFingerprint string

+ SHA256 string

+ Version int64

type Runner

+ func (r Runner) Lookup(ctx context.Context, s Spec) (Result, bool, error)

+ func (r Runner) RunWithResolver(ctx context.Context, s Spec, resolver ReuseResolver) (Result, error)

Oct 27, 2026

Changes in this version

type Runner

+ func (r Runner) RuntimeIdentity() string

Oct 23, 2026

Changes in this version

+ type BudgetDiagnostic struct

+ ConfiguredLimit int64

+ ContextWindow int64

+ CountBasis string

+ CountedInputTokens int64

+ CurrentUsage int64

+ EstimatedInputTokens int64

+ GenerationCalled bool

+ LimitKind string

+ MaxOutputTokens int64

+ Reason string

+ ReportedInputTokens int64

+ ReportedOutputTokens int64

+ RequestedReservation int64

+ Required int64

+ Unit string

+ func BudgetDiagnosticFromError(err error) *BudgetDiagnostic

+ type BudgetError struct

+ Diagnostic BudgetDiagnostic

+ func (e *BudgetError) Error() string

+ func (e *BudgetError) Unwrap() error

type HTTPProvider

+ func (p *HTTPProvider) InputTokenCountBasis() string

type OllamaProvider

+ func (p *OllamaProvider) InputTokenCountBasis() string

type OpenRouterProvider

+ func (p *OpenRouterProvider) InputTokenCountBasis() string

type Result

+ BudgetDiagnostic *BudgetDiagnostic

type SGLangProvider

+ func (p *SGLangProvider) InputTokenCountBasis() string

Oct 22, 2026

Changes in this version

type OllamaProvider

+ func (p *OllamaProvider) SelfHosted() bool

type SGLangProvider

+ func (p *SGLangProvider) SelfHosted() bool

+ type SelfHostedProvider interface

+ SelfHosted func() bool

type Spec

+ DailyTokenBudget *TokenBudget

+ type TokenBudget struct

+ InputTokens int64

+ OutputTokens int64

+ TotalTokens int64

Oct 20, 2026

Changes in this version

+ type OllamaOptions struct

+ FieldRoles map[string]string

+ Format string

+ GenerationTimeoutSeconds *int64

+ KeepAliveSeconds *int64

+ OutputTypes map[string]string

+ PromptMode string

+ Temperature *float64

+ Think *bool

type Request

+ Ollama *OllamaOptions

type Spec

+ Ollama *OllamaOptions

Oct 20, 2026

Changes in this version

+ func BoundedTextPrompt(req Request) string

Oct 16, 2026

Changes in this version

+ const SGLangRuntimeVersion

+ func SelfHosted(s Spec) bool

+ func ValidateSGLangConfig(cfg SGLangConfig) error

+ type OpenRouterConfig struct

+ CacheWriteUsage string

+ InputTokenUpperBound int64

+ UpstreamProvider string

+ type OpenRouterProvider struct

+ func NewOpenRouterProvider(apiKey, baseURL string, config OpenRouterConfig) (*OpenRouterProvider, error)

+ func (p *OpenRouterProvider) CountInputTokens(ctx context.Context, req Request) (int64, error)

+ func (p *OpenRouterProvider) Generate(ctx context.Context, req Request) (Response, error)

+ func (p *OpenRouterProvider) Identity() string

type Request

+ OpenRouter *OpenRouterConfig

+ SGLang *SGLangConfig

type Response

+ RuntimeCache *RuntimeCacheObservation

type Result

+ GenerationRuntimeCache *RuntimeCacheObservation

+ type RuntimeCacheObservation struct

+ CachedTokens *int64

+ DeviceTokens *int64

+ Evidence string

+ HostTokens *int64

+ Profile string

+ RuntimeVersion string

+ StorageBackend string

+ StorageTokens *int64

+ type SGLangCapabilities struct

+ GenerationSupported bool

+ Observed bool

+ Profile string

+ Reason string

+ RuntimeIdentitySHA256 string

+ RuntimeVersion string

+ type SGLangConfig struct

+ CacheNamespace string

+ DeploymentID string

+ MaxInputTokens int64

+ Profile string

+ SelfHosted bool

+ type SGLangProvider struct

+ func NewSGLangProviderContext(ctx context.Context, apiKey, baseURL, model string, cfg SGLangConfig) (*SGLangProvider, error)

+ func (p *SGLangProvider) Capabilities() SGLangCapabilities

+ func (p *SGLangProvider) CountInputTokens(ctx context.Context, req Request) (int64, error)

+ func (p *SGLangProvider) Generate(ctx context.Context, req Request) (Response, error)

+ func (p *SGLangProvider) Identity() string

+ func (p *SGLangProvider) RefreshIdentity(ctx context.Context) error

type Spec

+ OpenRouter *OpenRouterConfig

+ SGLang *SGLangConfig

Oct 14, 2026

Changes in this version

+ type GenerationTiming struct

+ Basis string

+ FirstTokenMS *float64

+ OutputMS *float64

+ Streaming bool

+ type Measurement struct

+ ContextBytes int

+ EndedAt time.Time

+ Fingerprint string

+ GenerationCalled bool

+ GenerationEndedAt time.Time

+ GenerationStartedAt time.Time

+ GenerationTiming *GenerationTiming

+ GenerationUsage *Usage

+ KnownGenerationUsage bool

+ OmittedContext int

+ Outcome string

+ PhaseMS map[string]float64

+ ProviderSetupEndedAt time.Time

+ ProviderSetupMS float64

+ ProviderSetupStartedAt time.Time

+ SelectedContext int

+ StartedAt time.Time

+ TotalContextBytes int

+ TotalMS float64

type OllamaProvider

+ func (p *OllamaProvider) ModelDigest() string

type Request

+ Stream bool

type Response

+ Timing *GenerationTiming

type Result

+ GenerationTiming *GenerationTiming

+ Measurement *Measurement

type Spec

+ Stream bool

Oct 13, 2026

Changes in this version

+ var ErrBudgetExceeded = errors.New("request exceeds the configured token or cost budget")

+ var ErrInFlight = errors.New(...)

+ var ErrIncomplete = errors.New(...)

+ type Broker interface

+ Get func(context.Context, string) (protocol.Artifact, error)

+ Publish func(context.Context, protocol.PublishRequest) (protocol.PublishResult, error)

+ type ContextItem struct

+ ID string

+ Priority int

+ Required bool

+ Stable bool

+ Text string

+ type HTTPProvider struct

+ func NewHTTPProvider(provider, apiKey, baseURL string) (*HTTPProvider, error)

+ func (p *HTTPProvider) CountInputTokens(ctx context.Context, req Request) (int64, error)

+ func (p *HTTPProvider) Generate(ctx context.Context, req Request) (Response, error)

+ func (p *HTTPProvider) Identity() string

+ type OllamaProvider struct

+ func NewOllamaProvider(baseURL, model string) (*OllamaProvider, error)

+ func NewOllamaProviderContext(ctx context.Context, baseURL, model string) (*OllamaProvider, error)

+ func (p *OllamaProvider) CountInputTokens(ctx context.Context, req Request) (int64, error)

+ func (p *OllamaProvider) Generate(ctx context.Context, req Request) (Response, error)

+ func (p *OllamaProvider) Identity() string

+ type Plan struct

+ ContextBytes int

+ Fingerprint string

+ Omitted []string

+ Request Request

+ Selected []string

+ TotalContextBytes int

+ func BuildPlan(s Spec) (Plan, error)

+ type Prices struct

+ CacheRead float64

+ CacheWrite float64

+ Input float64

+ Output float64

+ type Provider interface

+ CountInputTokens func(context.Context, Request) (int64, error)

+ Generate func(context.Context, Request) (Response, error)

+ Identity func() string

+ type Request struct

+ CacheMode string

+ ContextWindow int64

+ Input string

+ MaxOutputTokens int64

+ Model string

+ Prefix string

+ Provider string

+ ReasoningEffort string

+ type Response struct

+ Complete bool

+ ID string

+ Model string

+ StopReason string

+ Text string

+ Usage Usage

+ type Result struct

+ AvoidedCostUSD *float64

+ AvoidedUsage Usage

+ CostUSD *float64

+ CreatedAt time.Time

+ Durable bool

+ ExpiresAt time.Time

+ Fingerprint string

+ GenerationUsage Usage

+ Response Response

+ Reused bool

+ ThisRunUsage Usage

+ type Runner struct

+ Broker Broker

+ Now func() time.Time

+ Provider Provider

+ func (r Runner) Run(ctx context.Context, s Spec) (Result, error)

+ type Spec struct

+ CacheMode string

+ Context []ContextItem

+ DailyBudgetUSD float64

+ Input string

+ Instructions string

+ MaxContextBytes int

+ MaxCostUSD float64

+ MaxInputTokens int64

+ MaxOutputTokens int64

+ Model string

+ Prices *Prices

+ Provider string

+ ReasoningEffort string

+ ReuseForSeconds int64

+ Revision string

+ Task string

+ type Usage struct

+ CacheReadTokens int64

+ CacheWriteTokens int64

+ InputTokens int64

+ OutputTokens int64

+ ReasoningTokens int64

+ UncachedInputTokens int64