Struct PromptCachingConfig 

pub struct PromptCachingConfig {
    pub enabled: bool,
    pub cache_dir: String,
    pub max_entries: usize,
    pub max_age_days: u64,
    pub enable_auto_cleanup: bool,
    pub min_quality_threshold: f64,
    pub cache_friendly_prompt_shaping: bool,
    pub stable_tool_catalog_across_modes: bool,
    pub gap_warning_enabled: bool,
    pub gap_warning_threshold_secs: Option<u64>,
    pub providers: ProviderPromptCachingConfig,
}

Global prompt caching configuration loaded from vtcode.toml

Fields

enabled: bool

Enable prompt caching features globally

cache_dir: String

Base directory for local prompt cache storage (supports ~ expansion)

max_entries: usize

Maximum number of cached prompt entries to retain on disk

max_age_days: u64

Maximum age (in days) before cached entries are purged

enable_auto_cleanup: bool

Automatically evict stale entries on startup/shutdown

min_quality_threshold: f64

Minimum quality score required before persisting an entry

cache_friendly_prompt_shaping: bool

Enable prompt-shaping optimizations that improve provider-side cache locality. When enabled, VT Code keeps volatile runtime context at the end of prompt text.

stable_tool_catalog_across_modes: bool

Keep the wire tool catalog identical across planning and execution turns (union of both modes’ tools). Planning toggles then do not rewrite the tool prefix. The fail-closed execution gate still blocks mutations during planning.

gap_warning_enabled: bool

Warn before a request when the pause since the previous request likely exceeded the provider prompt cache lifetime (advisory only).

gap_warning_threshold_secs: Option<u64>

Override for the cache-gap warning threshold in seconds. When unset, a provider-specific default is used (Anthropic 300s, OpenAI 600s).

providers: ProviderPromptCachingConfig

Provider specific overrides

Implementations

impl PromptCachingConfig

pub fn resolve_cache_dir(&self, workspace_root: Option<&Path>) -> PathBuf

Resolve the configured cache directory to an absolute path

• ~ is expanded to the user’s home directory when available
• Relative paths are resolved against the provided workspace root when supplied
• Falls back to the configured string when neither applies

pub fn is_provider_enabled(&self, provider_name: &str) -> bool

Returns true when prompt caching is active for the given provider runtime name.

pub fn gap_threshold_secs(&self, provider_name: &str) -> Option<u64>

Resolve the cache-gap warning threshold in seconds for a provider.

Returns None when the warning should not fire: gap warnings disabled, caching disabled for the provider, or OpenAI configured with 24h extended retention (where a short pause cannot expire the cache).

Trait Implementations

impl Clone for PromptCachingConfig

fn clone(&self) -> PromptCachingConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for PromptCachingConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl Default for PromptCachingConfig

fn default() -> PromptCachingConfig

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for PromptCachingConfig

fn deserialize<__D>( __deserializer: __D, ) -> Result<PromptCachingConfig, <__D as Deserializer<'de>>::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl JsonSchema for PromptCachingConfig

fn schema_name() -> Cow<'static, str>

The name of the generated JSON Schema.

fn schema_id() -> Cow<'static, str>

Returns a string that uniquely identifies the schema produced by this type.

fn json_schema(generator: &mut SchemaGenerator) -> Schema

Generates a JSON Schema for this type.

fn inline_schema() -> bool

Whether JSON Schemas generated for this type should be included directly in parent schemas, rather than being re-used where possible using the $ref keyword.

impl Serialize for PromptCachingConfig

fn serialize<__S>( &self, __serializer: __S, ) -> Result<<__S as Serializer>::Ok, <__S as Serializer>::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.