Documentation
¶
Overview ¶
Package registry wires concrete LLM providers into an llmproxy.Client. This package is the only place that imports both llmproxy and provider/*; llmproxy itself has no knowledge of concrete provider implementations.
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func Build ¶
func Build(cfg ProviderConfig) []llmproxy.Provider
Build returns the set of providers enabled by the given configuration. Unconfigured providers are skipped; if a provider fails to initialize, it is logged and omitted rather than failing the whole build.
Types ¶
type EmbeddingConfig ¶
type EmbeddingConfig struct {
BatchSize int `env:"EMBEDDING_BATCH_SIZE" envDefault:"20"`
MaxConcurrent int `env:"EMBEDDING_MAX_CONCURRENT" envDefault:"1"`
MaxPerAccount int `env:"EMBEDDING_MAX_PER_ACCOUNT" envDefault:"3"`
BatchTimeout time.Duration `env:"EMBEDDING_BATCH_TIMEOUT" envDefault:"30s"`
QueryTimeout time.Duration `env:"EMBEDDING_QUERY_TIMEOUT" envDefault:"30s"`
ExtractTimeout time.Duration `env:"EMBEDDING_EXTRACT_TIMEOUT" envDefault:"5m"`
}
type ProviderConfig ¶
type ProviderConfig struct {
OpenAI openai.Config
Mistral mistral.Config
Gemini gemini.Config
Anthropic anthropic.Config
}
ProviderConfig holds the env-delivered API keys for the catalog LLM providers. Structured providers (self-hosted endpoints, backend-routed stand-ins) are not configured here.
Source Files
Click to show internal directories.
Click to hide internal directories.