framedeck-mcp
framedeck-mcp
Framedeck 的 MCP 服务器 — 将任何 AI 助手连接到您的内容生产流水线。
适用于 Claude、ChatGPT、Gemini、Copilot 以及任何兼容 MCP 的客户端。
快速入门
在 framedeck.app > 设置 > API 密钥处获取您的 API 密钥
添加到您的 AI 助手配置中(见下文)
Related MCP server: Calq MCP
设置
Claude Code
claude mcp add framedeck -e FRAMEDECK_API_KEY=your_key_here -- npx -y framedeck-mcpClaude Desktop
添加到您的 claude_desktop_config.json 中:
{
"mcpServers": {
"framedeck": {
"command": "npx",
"args": ["-y", "framedeck-mcp"],
"env": {
"FRAMEDECK_API_KEY": "your_key_here"
}
}
}
}其他 MCP 客户端
设置环境变量 FRAMEDECK_API_KEY 并运行:
npx framedeck-mcp使用示例
与您的 AI 助手自然对话 — 以下是三个端到端的工作流示例:
示例 1 — 捕捉并升级创意
您: “我刚才有了 3 个视频创意:磨刀教程、索尼 A7IV 评测和生活 Vlog。把它们加到我的创意池里。”
Claude (通过 MCP): 调用
create_multiple_cards并传入这三个标题。它们会进入您创意池的“创意”阶段。您: “磨刀那个可以做 — 把它升级为一个正式的生产项目。”
Claude: 调用
graduate_to_production,创建一个包含完整阶段(创意 → 脚本 → 拍摄 → 编辑 → 发布)的新生产项目,并将该创意卡片移动过去。
示例 2 — 规划生产日
您: “对于索尼评测,添加子任务:写开场白、拍摄产品镜头、录制旁白、剪辑 B-roll、调色、导出最终成品。”
Claude: 调用
set_active_board设置为“索尼 A7IV 评测”,然后调用create_multiple_cards在“拍摄”阶段创建所有六个子任务。您: “把开场白和产品镜头标记为已完成。”
Claude: 调用两次
move_card将它们移动到“已完成”阶段。
示例 3 — 获取项目概览
您: “我的频道进展如何?有什么逾期的吗?”
Claude: 为每个生产项目调用
get_board_overview。返回一个摘要,显示每个阶段的卡片数量、以红色高亮显示的逾期项目,以及哪些生产项目需要关注。您: “我现在被分配了什么任务?”
Claude: 调用
get_my_cards并按看板和状态列出您名下的所有任务。
工具 (32)
工具 | 描述 |
| 设置当前活跃的生产上下文 |
| 显示当前活跃的生产项目 |
| 列出所有生产项目/计划 |
| 创建一个新的生产项目或计划 |
| 列出看板中的阶段 |
| 创建一个新阶段 |
| 删除一个阶段 |
| 使用过滤器列出卡片/帧 |
| 创建一张卡片/帧 |
| 批量创建(最多 50 个) |
| 更新标题、描述、优先级、截止日期 |
| 移动到不同阶段 |
| 永久删除 |
| 软删除(可恢复) |
| 复制卡片(包含清单和标签) |
| 为卡片设置颜色编码 |
| 添加评论 |
| 添加子任务 |
| 勾选/取消勾选子任务 |
| 为卡片添加标签 |
| 移除标签 |
| 获取完整的卡片信息 |
| 按关键词搜索 |
| 指派团队成员 |
| 按状态分组获取我的卡片 |
| 获取包含逾期项目的看板摘要 |
| 列出所有标签 |
| 创建标签 |
| 将 git 提交关联到卡片 |
| 记录工作时间 |
| 活动摘要 |
| 将创意提升为完整生产项目 |
双模式
Framedeck 支持两种看板模式:
创作者模式:生产项目、阶段、帧 — 适用于内容流水线
经典模式:项目、阶段、卡片 — 适用于标准项目管理
MCP 服务器会根据每个看板的模式自动使用正确的术语。
创意池
当您在未指定看板的情况下添加创意时,它们会自动进入您的创意池。当创意准备就绪时,使用 graduate_to_production 创建一个包含完整阶段(创意 > 脚本 > 拍摄 > 编辑 > 发布)的专属生产项目。
安全注释
每个工具都附带 MCP 安全注释(readOnlyHint、destructiveHint、idempotentHint、openWorldHint),以便 AI 客户端在执行 delete_card、delete_column 和 archive_card 等破坏性操作前向用户发出警告。
隐私与支持
许可证
MIT
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
32 tool updates
v0.1.0- First observed
add_checklist_item - First observed
add_comment - First observed
add_labels_to_card - First observed
archive_card - First observed
assign_card - First observed
create_board - First observed
create_card - First observed
create_column - First observed
create_label - First observed
create_multiple_cards - First observed
delete_card - First observed
delete_column - First observed
duplicate_card - First observed
get_active_board - First observed
get_board_overview - First observed
get_card_details - First observed
get_my_cards - First observed
get_sprint_summary - First observed
graduate_to_production - First observed
link_commit - First observed
list_boards - First observed
list_cards - First observed
list_columns - First observed
list_labels - First observed
log_work - First observed
move_card - First observed
remove_label_from_card - First observed
search_cards - First observed
set_active_board - First observed
set_card_color - First observed
toggle_checklist_item - First observed
update_card
TDQS
Scored across 32 tools
Tools are semantically distinct with clear resource targeting (card vs board vs column vs label) and action specificity (create vs duplicate vs graduate, add vs toggle). While the large set creates cognitive load, descriptive prefixes prevent overlap between similar operations like list_cards (filtered) and search_cards (keyword across all boards).
Exemplary consistency across all 32 tools. Every name follows snake_case with standardized verb_object structure (e.g., create_card, add_labels_to_card, set_active_board, toggle_checklist_item). Verbs are uniform throughout the CRUD and lifecycle operations.
32 tools significantly exceeds the threshold for effective agent selection (25+). The surface is overly granular, exposing fine-grained operations (separate tools for color, checklist items, single vs batch create) that could be consolidated into parameters, increasing selection error rates.
Strong coverage of the project management domain including full card lifecycle, board/column management, labeling, assignment, time tracking, and git integration. Minor gaps exist (no delete board, no update/delete comment), but core video production workflows (graduate_to_production, Idea Pool) are well supported.
Maintenance
Related MCP Connectors
Collaborative whiteboard MCP server — create objects, connectors, C4 diagrams, and manage boards
MCP server for progressive tool usage at any scale (see https://klavis.ai)
MCP server for Product Management
Hosted MCP server for AI-driven data ops. Create apps, manage schemas, and CRUD structured data.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceA comprehensive Model Context Protocol (MCP) server for Notion integration with enhanced functionality, robust error handling, production-ready features, and bulletproof validation.5-
- FlicenseNot gradedqualityDmaintenanceA comprehensive MCP server for time tracking, project management, and AI-powered memory storage using semantic search. It enables users to log time, manage client billing, and capture shared or personal ideas through integrated tools and team collaboration features.1-
- AlicenseAqualityCmaintenanceMCP server for Fizzy kanban task management with tools for boards, cards, comments, and checklists.725 npm3AGPL 3.0
- AlicenseAqualityCmaintenanceMCP server for YouTube creator-ops — video metadata, comments, playlists, channel analytics, plus a ComfyUI bridge for AI thumbnail generation.1546 npm1MIT