plugin.json

{
  "$schema": "https://agent-plugins.org/schemas/1.0.0/plugin.schema.json",
  "name": "convosight-scp",
  "version": "1.0.0",
  "description": "Ask Convosight's AI audience personas for their opinions and reactions to ads, messaging, products and ideas.",
  "author": {
    "name": "Convosight",
    "email": "reachus@convosight.com",
    "url": "https://convosight.com"
  },
  "homepage": "https://convosight.com",
  "keywords": [
    "personas",
    "audience",
    "research",
    "marketing",
    "digital twin",
    "consumer insights"
  ],
  "extensions": {
    "com.openai": {
      "interface": {
        "displayName": "Persona",
        "shortDescription": "Chat with consumer personas",
        "longDescription": "Convosight's Persona gives you simulated consumer personas built from real social and consumer research. Browse the personas and ask them questions to get realistic opinions and reactions to ads, messaging, products and ideas. Pick a persona, ask your question, and read the answer in an interactive chat widget. Open the full report for any conversation. Limitations: answers are AI-simulated persona responses, not real survey data, and a daily question limit applies per user. Requires signing in with a Convosight account.",
        "developerName": "Convosight",
        "category": "Productivity",
        "capabilities": [
          "Read",
          "Write"
        ],
        "websiteURL": "https://convosight.com",
        "supportURL": "https://convosight.com/mcp-support/faqs",
        "privacyPolicyURL": "https://convosight.com/policies/privacy",
        "termsOfServiceURL": "https://convosight.com/policies/tos",
        "defaultPrompt": [
          "Ask a persona what they think of my new ad headline.",
          "Show me the available personas.",
          "How often do you use a specific skincare product?"
        ],
        "brandColor": "#10A37F",
        "composerIcon": "./assets/composer-icon.png",
        "logo": "./assets/logo.png",
        "screenshots": [
          "./assets/screenshot-1.png",
          "./assets/screenshot-2.png",
          "./assets/screenshot-3.png"
        ]
      },
      "review": {
        "test_cases": {
          "positive": [
            {
              "description": "In a new chat, the user wants a consumer's reaction to a product but hasn't named a persona. They pick a persona in the widget (e.g. Kutaiba Ziad, 18, Saudi Arabia, hair care) and the question is sent to that persona.",
              "prompt": "Ask a persona: What do you think about this hair oil? It's a rosemary and castor oil blend, non-greasy, 100 ml for 45 SAR.",
              "tools_triggered": "list_personas (with question = the user's message); after the user picks a persona, the widget calls ask_persona_widget and get_persona_answer itself",
              "expected_behavior": "The persona picker opens and ChatGPT stops. After the user picks Kutaiba Ziad, the widget sends the message unchanged and streams her first-person reaction like a real consumer would give it: what appeals to her, what she doubts (e.g. whether rosemary really works, greasiness, price), and whether she would try it. A \"View full report\" link is shown. ChatGPT does not add its own opinion."
            },
            {
              "description": "In a new chat, the user names a persona and asks about their own problems and habits in the category, the way you would interview a real consumer.",
              "prompt": "Ask Kutaiba Ziad: What kind of hair problems do you face, and what are you using for them right now?",
              "tools_triggered": "list_personas (search = \"Kutaiba Ziad\", question = the user's question)",
              "expected_behavior": "Exactly one persona matches, so the chat starts immediately without the picker. Kutaiba answers in first person about her own hair concerns (e.g. visible hair fall or thinning) and the products and home remedies she currently uses, consistent with her profile. The question is passed verbatim."
            },
            {
              "description": "Run right after the previous case in the same chat. The user asks a follow-up about satisfaction with her routine without naming the persona; it must go to the same persona and session.",
              "prompt": "Are you satisfied with your current routine? What would make you switch to something else?",
              "tools_triggered": "ask_persona (same persona_id and session_id as the previous case; list_personas is not called)",
              "expected_behavior": "Kutaiba replies in the Persona chat widget, building on her previous answer: how satisfied she is with what she uses now, what frustrates her, and what would make her switch (e.g. visible results, trusted reviews, price). The picker does not reappear and ChatGPT does not answer for her."
            },
            {
              "description": "In the same chat, the user explicitly switches to a persona in a different category and country and asks about his everyday habits.",
              "prompt": "Switch to Leland Bishop and ask him: What drinks do you usually grab during the day, and why those?",
              "tools_triggered": "list_personas (search = \"Leland Bishop\", question = the user's question), starting a new session for Leland",
              "expected_behavior": "A new chat starts with Leland Bishop (24, USA, ready-to-drink beverage consumer) under a new session_id. He answers in first person about his daily drink choices and the reasons behind them (e.g. bold flavors, seasonal drops, what's trending). Kutaiba's earlier context does not carry over."
            },
            {
              "description": "Run right after the previous case in the same chat. The user shows Leland a product concept and asks for his honest reaction and purchase intent.",
              "prompt": "What do you think about a limited-edition mango chili iced tea in a can for $4? Would you buy it?",
              "tools_triggered": "ask_persona (same persona_id and session_id as the previous case; list_personas is not called)",
              "expected_behavior": "Leland replies in the Persona chat widget with his honest first-person reaction to the concept and the $4 price, and a clear yes, no or maybe on buying it with his reasons. He stays consistent with his earlier answer. \"View full report\" opens the full conversation report."
            }
          ],
          "negative": [
            {
              "description": "User asks for general market-size data about a consumer category. This is factual research, not a request for a persona's opinion, so the plugin should not be invoked. Expected: ChatGPT answers normally (e.g. from its own knowledge or web search) without calling list_personas or ask_persona.",
              "prompt": "What's the market size of ready-to-drink iced tea in the US?"
            },
            {
              "description": "User asks ChatGPT to write marketing copy. This is content creation, not a request for a persona's reaction, so the plugin should not be invoked. Expected: ChatGPT writes the headlines itself without calling any Persona tool.",
              "prompt": "Write three ad headlines for a summer iced tea launch."
            },
            {
              "description": "User wants ChatGPT to draft a fictional buyer-persona document for a marketing plan. They are not asking to talk to a Convosight persona, so the plugin should not be invoked. Expected: ChatGPT drafts the profile itself without opening the persona picker.",
              "prompt": "Create a buyer persona profile for a 30-year-old fitness enthusiast for my marketing plan."
            }
          ]
        },
        "demo_recording_url": "https://www.loom.com/share/30d337a5c7d34ed9a19b0368088213a0"
      },
      "publication": {
        "release_notes": "Persona by Convosight, v1.0.0 (initial release)\n\nChat with simulated consumer personas, right inside ChatGPT.\n\nPersona gives you AI consumer personas built from real consumer conversations on social media: posts, videos and comments across platforms like TikTok, Instagram, YouTube, Reddit and Facebook. Convosight analyzes these conversations to learn how real consumers talk about categories, brands and products. Each persona brings those patterns to life as one believable consumer. Ask a persona about their habits, problems and routines, or what they think of your product, ad or idea, and get a realistic answer in their own voice.\n\nWhat's included\n- Persona browser. Browse the available personas, or ask a question and pick the persona that fits from an interactive widget.\n- Persona chat. Talk to the persona you picked. Follow-up questions stay with the same persona until you ask to switch.\n- Interactive chat widget. Read and continue the conversation in a rich in-chat view.\n- View full report. Open the full report for any conversation.\n\nSign-in and access\n- Requires signing in with a Convosight account.\n- A usage limit applies per user. You're told plainly when you reach it, with the reset time when one is available.\n\nLimitations\n- Answers are AI-simulated persona responses based on patterns in public social conversations. They are not quotes from real individuals, and they are not survey data.\n- Best for qualitative reactions and opinions, not general research, analytics or factual lookups.\n\nSupport: https://convosight.com/mcp-support/faqs"
      },
      "apps": "./.app.json"
    }
  }
}