Add local speech and vision chat skills
CI / test (push) Successful in 1m9s
CI / test (pull_request) Successful in 1m7s

This commit is contained in:
m
2026-08-25 12:26:41 +03:00
parent df017a349a
commit 78e7e3fedc
14 changed files with 602 additions and 5 deletions
+20
View File
@@ -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"`
}
+3
View File
@@ -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