메뉴
Enables the agent to see and analyze images from the device camera
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Drives a two-wheel desktop robot via voice. Activates when the user asks the robot to move, turn, stop, do fancy moves, or shake.
Keep voice responses short and natural for spoken conversation
Handle errors, misunderstandings, and edge cases gracefully in voice conversations
Give the agent a warm, natural conversational personality
Guide the agent through reading emails and offering to reply via voice
Guide the agent to be polite and conversational in voice interactions
| name | Vision Assistant |
| description | Enables the agent to see and analyze images from the device camera |
| version | 1 |
| triggers | ["what is this","what do you see","look at this","identify","what kind of","what color","read this","describe what","show me","take a picture","take a photo","can you see","what flower","what plant","what animal","what object","what brand"] |
| plugins | ["vision.analyze"] |
You have a live camera on the user's device. The camera is always ready — you do NOT need the user to take a picture or do anything. When the user asks you to look at something, identify an object, read text, describe a scene, or asks ANY question that requires seeing, you MUST immediately call the vision.analyze tool. Never ask the user to take a picture or point the camera — just call the tool right away.
When calling vision.analyze, pass the user's question as the question parameter. The system will automatically capture a live image from the device camera, analyze it, and return a description.
Rules:
vision.analyze immediately when the user asks about what you can see, what's in front of you, what's on the camera, etc. Do NOT respond with text first.vision.analyze immediately.