Flow Relay MCP Server
Enables AI assistants to manage Flow Relay projects, generate handoffs and insights, and interact with integrations like Discord through natural language.
README
Flow Relay MCP Server
Flow Relay MCP Server adds project-aware, multi-tenant Flow Relay tools to MCP clients such as Claude Desktop and Claude Code.
It connects to Flow Relay API v1 using an API key and supports:
- Project scope (personal project or organization project). Every handoff and AI insight is tied to a project. Events and integrations remain user-level.
Package
- Name: @flowrelay/mcp-server
- Version: 1.0.14
What Is Included
The server currently exposes these tools:
- get_workspace_context
- list_projects
- set_active_project
- list_filter_options (lists the selectable resources, branches, event types and priorities per source for a project – call this before generating so handoff/insight filters use real values instead of guesses; also reports whether Figma is selectable, since Figma requires the project's processing region to be Global)
- list_handoffs (returns project-specific handoffs, or an aggregated view of all accessible project handoffs if no project is active; filter by status: active, archived, or all)
- ask_project (ask a question about a project and get an answer grounded in codebase, baselines and 14 days of activity; 2 credits per question)
- generate_handoff (processed asynchronously, and the server automatically polls until the job finishes. Requires active project or project_id. Accepts per-source filters: projects, eventTypes, branches, priorities.)
- generate_correlation_insight (accepts per-source filters)
- generate_onboarding_brief (accepts per-source filters)
- generate_architecture_insight (accepts per-source filters)
- generate_release_notes (generates release notes or PR descriptions from code activity; accepts per-source filters)
- list_digests (lists past scheduled activity digests for a project)
- list_insights
- list_integrations
- list_events
- list_untracked_resources (lists event-producing resources not scoped to any project – useful for discovering untracked activity)
- discord_list_channels
- discord_send_message (plain text, or a handoff/insight by id or
last_*–last_handoff,last_correlation,last_onboarding,last_architecture,last_release_notes– rendered to Markdown and attached as a.mdfile)
Environment Variables
Required:
- FLOWRELAY_API_KEY
Optional:
- FLOWRELAY_PROJECT_ID
- FLOWRELAY_BASE_URL
If FLOWRELAY_PROJECT_ID is set, it becomes the default project context for project-aware tools unless you override it per call.
Quick Start (Claude Desktop)
Add this to your claude_desktop_config.json:
{
"mcpServers": {
"flowrelay": {
"command": "npx",
"args": ["-y", "@flowrelay/mcp-server"],
"env": {
"FLOWRELAY_API_KEY": "fr_your_api_key_here",
"FLOWRELAY_PROJECT_ID": "your_project_id"
}
}
}
}
Multi-Tenant Behavior
- Every handoff and insight is tied to a project; you need an active project (
set_active_project) orproject_id. Events and integrations remain user-level. - You can select a project during the MCP session with set_active_project.
- You can override scope per call by passing project_id where supported.
Recommended flow:
- Call get_workspace_context
- Call list_projects
- Call set_active_project
- Run handoff and query tools in the selected scope
Local Development
From this folder:
npm install
npm run build
Create a tarball package:
npm pack
Troubleshooting
- Error: Missing FLOWRELAY_API_KEY
- Set FLOWRELAY_API_KEY in your MCP client configuration.
- Project not found or inaccessible
- Run list_projects and use one of the returned IDs.
- No events or handoffs returned
- Verify active scope and data availability in that scope.
Related Docs
- Documentation & Platform: https://www.flowrelay.it
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。