vibex-workflow-mcp

vibex-workflow-mcp

Enables creating, inspecting, validating, publishing, and debugging VibeX DAG Workflows as source files, with run control and safe revision management.

Category
Visit Server

README


summary: Create, inspect, debug, and safely revise VibeX DAG Workflows from Agent conversations and editable file tabs.

Workflow Creator

Author VibeX DAG workflows as real files, then validate, debug, and publish them without treating an in-memory graph as the source of truth.

Once enabled, three surfaces work together:

  • the vibex-workflow-creator Skill, so an Agent can design a valid DAG;
  • vibex-workflow-mcp, for read/write, validation, publish, run inspection, pause/continue, and debug runs;
  • Workflow Studio, which opens *.vibex-workflow.json as an editable tab.

The source file is the authoring fact. Saving uses an artifact revision and refuses a stale overwrite. Publishing creates an immutable version. Automations stay on the version they already reference until you point them at a new one.

Enable

This plugin ships with the Host, starts disabled, and cannot be uninstalled. Turn it on in Plugins. The first enable binds the Skill and MCP to compatible installed Agents. Later enable/disable cycles keep the selection you already made.

VibeX Desktop must be running. The Host starts a loopback-only command gateway, launches the MCP with the managed Node.js runtime, and injects a per-app-lifetime bearer token. That token is never stored in this package or in config.json.

Edit a workflow

Create or open a *.vibex-workflow.json file in the project, or under ~/.vibex/workflows/. The tab is Workflow Studio. Saving writes the source artifact; it does not publish a version and does not retarget Automations.

Model nodes as durable steps, not as “the Agent”. An Agent Step owns one child Conversation. Retries and continuation add Turns to that same Conversation. Encode order only through dependsOn. Do not create cycles.

Debug and publish

Save and validate before any test.

  • workflow_debug_source runs the current unpublished source through a hidden durable snapshot. It must not create a catalog version.
  • workflow_publish creates an immutable visible version, and only when you ask for a release.
  • workflow_start starts a published version against a real workspace.
  • Pause stops active Turns without marking the run failed. Continue creates a new Turn in the same node Conversation.
  • A derived debug run can reuse unchanged completed ancestors and rerun one node, or that node plus its downstream.

Configuration

Default Agent Step completion chooses how a new Agent Step finishes if the source does not say otherwise:

  • Manual: wait for confirmation after the step produces a candidate.
  • Automatic: accept the candidate and continue downstream.

This is an authoring default. It does not rewrite steps that already declare a completion policy, and it does not change runs that already started.

Offline and network

The editor and MCP stay on the machine. MCP traffic uses a random loopback port between the managed process and VibeX. Agent Steps follow the selected Agent and project network policy; the plugin does not add its own outbound access.

Troubleshooting

  • MCP says VibeX is unavailable: keep Desktop open and re-enable the plugin so the Host can inject a fresh token.
  • Save conflict: the source changed outside the tab. Reload, reconcile, then save with the new revision.
  • An Automation did not pick up edits: it is pinned to one published version. Publish, then point that Automation at the new version.
  • Debug created a catalog version: stop. Source tests must use workflow_debug_source, not publish.

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured
E2B

E2B

Using MCP to run code via e2b.

Official
Featured