FunTell Project Management

You have access to the FunTell MCP server tools. Use them to turn any information into AI-powered content experiences. Projects can cover products, venues (museums, restaurants, events), education (courses, tutorials), knowledge bases, portfolios, storytelling, print materials, and more. Audiences access content via QR codes, direct links, or embeds.

Authentication

Always start by calling login with the creator's FunTell platform credentials (email + password). All other tools require authentication.

After login, optionally call get_subscription to check the creator's tier and limits:

• •Free: 3 projects, 50 sessions/month, no translations
• •Starter: 5 projects, $40 session budget, max 2 translation languages
• •Premium: 35 projects, $280 session budget, unlimited translations

Three Creation Modes

Choose the mode based on the user's input.

Mode A: Document with images (PDF/Word → archive ZIP)

Use when: The user uploads a PDF or Word document (.docx) that contains embedded images they want included in their project.

MCP tools cannot handle images. This mode runs a local script to extract text + images into a ZIP archive, which the user imports via the web portal.

Steps:

1. •

   Check dependencies are installed:

   bash

   python3 -c "import fitz; import docx" 2>/dev/null || pip3 install -r funtell-plugin/skills/manage/scripts/requirements-doc.txt
   

2. •

   Determine the best options from the document content:

   • •--name: Project name (ask user or infer from document title)
   • •--mode: list / grid / cards / single (use Content Mode Decision Guide below)
   • •-g: Add if document has clear sections/categories
   • •--language: Detect from document text (default: en)
3. •

   Run the script:

   bash

   python3 funtell-plugin/skills/manage/scripts/doc-to-archive.py <file_path> --name "Project Name" --mode list -g
   

4. •

   Check the summary output. If category/item count looks wrong, unzip, edit project.json, re-zip.

5. •

   Tell the user:

   Your archive is ready at {output_path}. To import:

   1. •Open the FunTell web portal
   2. •Go to Dashboard > Import
   3. •Upload the ZIP file
   4. •Review and confirm the import

6. •

   After import, the user can configure AI settings via the web portal or ask you to use generate_ai_settings + update_project MCP tools.

For archive format details, see scripts/format-reference.md.

Mode B: Document without images (text-only → MCP tools)

Use when: The user uploads a PDF, Word doc, or other file and doesn't need images included, or the document has no meaningful images.

This mode is faster — it creates the project directly via MCP tools with no manual import step.

Steps:

1. •Read and extract all text content from the document.
2. •Identify structure — sections, headings, chapters, categories, items.
3. •Automatically determine settings (content_mode, is_grouped, categories, original_language).
4. •Create project and content via MCP tools (see Project Setup Sequence below).
5. •Generate AI settings and enable AI.
6. •Present summary. Remind user to upload images and add translations via the web portal.

Key principles:

• •Maximize automation — make all decisions based on the document content.
• •Write rich, detailed descriptions for each item (directly impacts AI quality).
• •Use optimize_description to polish rough descriptions.
• •Present your decisions and ask for confirmation if the structure is ambiguous.

Mode C: Consultative (interactive Q&A)

Use when: The user describes what they want without providing source material.

Question flow:

1. •Purpose: "What is this project about?" — Understand the use case.
2. •Content mode: Recommend a content_mode and is_grouped setting with reasoning, but present alternatives. Example: "I recommend list mode with grouped categories for your restaurant menu. Alternatively, grid mode works well if you have photos for every dish. Which do you prefer?"
3. •Structure: Suggest category names, let user confirm or adjust.
4. •Content items: Gather names and descriptions, or offer to generate from brief notes.
5. •Confirm all settings before creating.
6. •Create everything via MCP tools (see Project Setup Sequence below).

Key principles:

• •Present recommendations with reasoning, but let the user decide.
• •Never silently decide project settings.
• •Ask focused, one-topic-at-a-time questions.

If unclear which mode the user prefers, ask: "Would you like to upload a document, or shall I walk you through setting up the project step by step?"

Content Mode Decision Guide

Project Setup Sequence (for Modes B and C)

1. Create the project

Call create_project with name, content_mode, is_grouped, original_language, billing_type. Leave conversation_ai_enabled: false — enable after AI setup.

2. Add content

Call create_content_items_batch with categories and items:

• •Non-grouped (is_grouped: false): Single category named "default"
• •Grouped (is_grouped: true): Multiple named categories

Each item needs: name, description (detailed markdown), optional ai_knowledge_base (max 500 words).

Markdown Formatting Rules

Content renders on a mobile dark-background UI:

Use: paragraphs, bold, italic, ###/#### headings, bullet lists, numbered lists, [links](url), > blockquotes, --- rules

Do NOT use: tables, code blocks, HTML tags, h1/h2 headings, nested lists beyond 2 levels

Rules:

• •Do NOT start with the item name (already displayed as title)
• •Short paragraphs (2-4 sentences) for mobile readability
• •Bold for key terms, prices, highlights
• •Bullet lists for structured info (specs, features, hours)
• •For tabular data, use labeled bullets instead of tables

3. Generate AI settings

Call generate_ai_settings with project ID and name. Returns ai_instruction, ai_knowledge_base, ai_welcome_general, ai_welcome_item.

4. Enable AI assistant

Call update_project with all four AI settings + conversation_ai_enabled: true.

5. Verify content

list_content_items to review, then adjust with reorder_content_item, move_content_item, update_content_item, delete_content_item.

6. Manage distribution

Default QR code is auto-created. Use list_access_tokens, create_access_token, update_access_token, delete_access_token.

7. Remind user about remaining tasks

• •Upload images (if not already included via Mode A)
• •Add translations if multilingual content is needed (supports: en, zh-Hant, zh-Hans, ja, ko, es, fr, ru, ar, th)

For detailed tool parameters, see reference.md.

Ongoing Management

• •list_projects — overview of all projects with stats
• •get_project — full project details
• •get_credit_balance — monitor remaining credits
• •optimize_description — AI-polish any description

Tips

• •Description quality matters: Detailed descriptions directly improve AI assistant responses.
• •Batch creation is preferred: Use create_content_items_batch over individual calls.
• •Check subscription first: Session limits depend on the creator's tier.
• •Supported languages: en, zh-Hant, zh-Hans, ja, ko, es, fr, ru, ar, th