flow-mcp
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@flow-mcp pause the current workflow"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
flow-control-mcp
极简、稳定的业务控制原语 MCP Server。
flow-control-mcp 只做一件事:把 Agent/Skill 调用的 flow.* 业务控制工具,写成 .runs/state.json.businessControl 中可验证、可审计、可调度消费的结构化事实。
1. Install
npm install -g flow-control-mcp本地开发:
npm install
npm run buildNode.js 要求 >=20。
Related MCP server: relaycore
2. Generate Claude Code Config
flow-control-mcp config claude --json输出可作为跨用户共享模板:
{
"mcpServers": {
"flow-control-mcp": {
"command": "flow-control-mcp",
"args": [],
"env": {
"FLOW_MCP_REQ_PATH": "${FLOW_MCP_REQ_PATH}"
}
}
}
}FLOW_MCP_REQ_PATH 应由 Agent Runtime 按会话注入;不同会话可以注入不同 request path。
前提:目标环境已通过 npm install -g flow-control-mcp 安装,并且 flow-control-mcp 在 Runtime PATH 中可用。Server 启动时仍会校验 Node.js >=20。
3. Verify
flow-control-mcp doctor --req-path /absolute/request/path --jsondoctor 会检查实际 Node 路径、Node 版本、.runs、control-context.json、state.json、写权限和锁。
开发门禁:
npm run typecheck
npm run build
npm test
npm run test:integration
npm pack --dry-run打包:
# 默认:不包含 runtime dependencies,安装时由 npm 拉取依赖
npm run pack:default
# 离线部署:包含 runtime dependencies
npm run pack:bundledContract
Server 名称:
flow-control-mcpTools:
flow.finish、flow.pause、flow.resume、flow.rewind、flow.abortContext:
FLOW_MCP_REQ_PATH/.runs/control-context.jsonState:
FLOW_MCP_REQ_PATH/.runs/state.json.businessControlLogs:运行日志只写 stderr;stdout 只用于 MCP 协议或 CLI 输出
最小 control-context.json:
{
"runId": "run-123",
"stage": "design",
"actionIndex": 1
}更多 Claude Code、Linux、Windows 和故障定位说明见 docs/claude-code.md。
Skill 中如何调用 5 个业务控制原语见 docs/skill-usage.md。
Available Tools
5 toolsflow.abortAbort business flowCDestructive
Terminate the current business flow without treating it as business failure.
Inputs: reason (required), message. Returns: ok, server, command, state, +6 more
flow.finishFinish business executionADestructive
End the current business action with an explicit success or failure conclusion.
Inputs: result (required), reason, message. Returns: ok, server, command, state, +6 more
flow.pausePause business executionADestructive
Pause the current business action while waiting for a user, dependency, or external system.
Inputs: reason (required), message. Returns: ok, server, command, state, +6 more
flow.resumeResume business executionBDestructive
Mark a waiting business action as ready for continued scheduling.
Inputs: message. Returns: ok, server, command, state, +6 more
flow.rewindRequest business rewindCDestructive
Request a rewind to the last stable point or an authorized checkpoint before re-execution.
Inputs: rewindTo (required), reason (required), message. Returns: ok, server, command, state, +6 more
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 5 tools
Each tool targets a distinct action in the flow lifecycle: abort (terminate without failure), finish (end with success/failure), pause (wait), resume (continue), rewind (go back to checkpoint). No overlap.
All tool names are single lowercase verbs (flow.abort, flow.finish, etc.), following a consistent verb-only pattern that describes the action clearly.
Five tools is appropriate for managing business flow lifecycle. It covers the essential operations without being excessive or too sparse.
The set covers key flow actions (abort, finish, pause, resume, rewind) but is missing a 'start' or 'initiate' tool, assuming flows are started externally. Minor gap for a complete lifecycle.
Maintenance
Related MCP Connectors
Hosted MCP memory and agent control plane for durable conversations, jobs, and operations.
Guarded MCP server for agent-readable business truth, provenance, readiness, and discovery.
Control plane for autonomous software labor. Agents claim objectives over MCP with audit trail.
Build and run grounded business agents over MCP: agents, knowledge bases, skills, Storylines.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceMCP server orchestrating local multi-agent workflows with gated lifecycle, handoff events, and host-level continuation.-
- AlicenseNot gradedqualityCmaintenanceA lightweight, self-hostable MCP server for shared memory, structured command relay, and traceable decision evidence across AI runtimes.1MIT
- AlicenseNot gradedqualityCmaintenanceHarness-agnostic MCP server for orchestrating deterministic, resumable multi-agent workflows. It lets you start, watch, resume, and cancel agent runs from any MCP client, using agents from Codex, Claude, Cursor, or Kimi.5MIT
- AlicenseAqualityCmaintenanceMCP server that enables running and managing file-based AI Skills, Agents, and Flow workflows as DAGs, with tools for listing, executing, and resuming tasks via stdio JSON-RPC.9MIT