Add local speech and vision chat skills
This commit is contained in:
@@ -0,0 +1,20 @@
|
||||
package capability
|
||||
|
||||
// MultimodalOperation identifies one bounded local media skill. It is kept
|
||||
// separate from general cognition so health, routing and failure evidence do
|
||||
// not collapse into a generic model response.
|
||||
type MultimodalOperation string
|
||||
|
||||
const (
|
||||
MultimodalTranscribe MultimodalOperation = "TRANSCRIBE_AUDIO"
|
||||
MultimodalUnderstand MultimodalOperation = "UNDERSTAND_IMAGE"
|
||||
MultimodalRecognizeOCR MultimodalOperation = "RECOGNIZE_OCR"
|
||||
)
|
||||
|
||||
type MultimodalResponse struct {
|
||||
Operation MultimodalOperation `json:"operation"`
|
||||
Text string `json:"text"`
|
||||
Provider string `json:"provider,omitempty"`
|
||||
ObservedAt int64 `json:"observed_at_unix_millis"`
|
||||
ConfidenceNote string `json:"confidence_note,omitempty"`
|
||||
}
|
||||
@@ -29,6 +29,9 @@ const (
|
||||
CapabilityKindCognitiveReasoning CapabilityKind = "COGNITIVE_REASONING"
|
||||
CapabilityKindScheduleReminder CapabilityKind = "SCHEDULE_REMINDER"
|
||||
CapabilityKindVerifyFact CapabilityKind = "VERIFY_FACT"
|
||||
CapabilityKindSpeechToText CapabilityKind = "SPEECH_TO_TEXT"
|
||||
CapabilityKindImageUnderstanding CapabilityKind = "IMAGE_UNDERSTANDING"
|
||||
CapabilityKindOCR CapabilityKind = "OCR"
|
||||
)
|
||||
|
||||
// Binding is Capability-owned correctness state. SecretRef is an opaque
|
||||
|
||||
Reference in New Issue
Block a user