Documentation
¶
Overview ¶
Package tool provides high-level utilities for common AI workflows. It simplifies single-shot "tool call" patterns where a consistent action is applied to varying inputs.
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type Config ¶
type Config struct {
// SystemPrompt defines the tool's behavior and role.
// This is cached automatically when the tool is reused.
SystemPrompt string
// MaxTokens is the maximum number of tokens to generate.
// Defaults to 4096 if not specified.
MaxTokens int
// Temperature controls randomness (0.0 = deterministic, 1.0 = creative).
// Defaults to 0.7 if not specified.
Temperature float64
// CacheTTL is the cache time-to-live for the system prompt.
// Valid values: "5m" (5 minutes) or "1h" (1 hour).
// Defaults to "5m" if not specified.
CacheTTL string
}
Config holds configuration for a Tool instance. All fields are optional and will use sensible defaults if not specified.
type Tool ¶
type Tool struct {
// contains filtered or unexported fields
}
Tool represents a reusable AI tool with consistent behavior. Create a Tool instance with New(), then call Run() multiple times with different inputs. The system prompt is cached for efficiency.
func New ¶
New creates a new Tool instance with the given configuration. The configuration is applied once and reused for all subsequent Run() calls.
func (*Tool) Run ¶
Run executes the tool on the given input string. The input is wrapped in XML document tags to provide clear semantic separation from the system prompt. This is especially important for inputs that might contain confusing content (like conversation logs with user/assistant markers).
The system prompt is cached automatically for efficiency when the tool is reused multiple times.
func (*Tool) RunWithResponse ¶
RunWithResponse executes the tool and returns the full completion response. This includes token usage information for cost tracking.
Source Files