optimizer

Documentation

Overview

Package optimizer uses an OpenAI-compatible LLM endpoint to merge, deduplicate, and consolidate multiple instruction files into a single coherent output.

Index

• func ConcatRaw(instructions []ContentInput) string
• type ContentInput
• type LLMConfig
• type Optimizer
  • func New(cfg LLMConfig, client ...*http.Client) *Optimizer
  • func (o *Optimizer) Enabled() bool
  • func (o *Optimizer) Optimize(ctx context.Context, instructions []ContentInput) (string, error)

Functions

func ConcatRaw

func ConcatRaw(instructions []ContentInput) string

ConcatRaw is the fallback when LLM is not available — simple concatenation.

Types

type ContentInput

type ContentInput struct {
	Source  string
	Path    string
	Content string
}

ContentInput is the input to the optimizer.

type LLMConfig

type LLMConfig struct {
	// Endpoint is the base URL (e.g. "https://api.openai.com/v1").
	Endpoint string `yaml:"endpoint"`
	// Model is the model name (e.g. "gpt-4o-mini").
	Model string `yaml:"model"`
	// APIKey is the bearer token.
	APIKey string `yaml:"-"`
	// Enabled controls whether LLM optimization is on by default.
	Enabled bool `yaml:"enabled"`
}

LLMConfig configures the OpenAI-compatible LLM used to consolidate content.

type Optimizer

type Optimizer struct {
	// contains filtered or unexported fields
}

Optimizer calls an OpenAI-compatible chat completions endpoint to consolidate content.

func New

func New(cfg LLMConfig, client ...*http.Client) *Optimizer

New creates an Optimizer. Returns nil if LLM is not configured. An optional *http.Client can be passed to control proxy/TLS behavior; if nil, a default client with a 60 s timeout is used.

func (*Optimizer) Enabled

func (o *Optimizer) Enabled() bool

Enabled returns true if the optimizer is configured and available.

func (*Optimizer) Optimize

func (o *Optimizer) Optimize(ctx context.Context, instructions []ContentInput) (string, error)

Optimize takes multiple instruction contents and returns a single consolidated version.