| icon | server |
|---|---|
| description | Where to start with the vCon MCP Server, the open-source Model Context Protocol server that lets AI assistants query and update a store of vCons. |
The vCon MCP Server gives an AI assistant or agent access to a store of vCons over the Model Context Protocol. It runs over stdio or Streamable HTTP, exposes 46 tools in seven groups plus resources and prompts, and stores vCons in Supabase Postgres with optional Redis caching and pgvector semantic search.
The code is at vcon-dev/vcon-mcp, MIT-licensed TypeScript,
published as the npm package vcon-mcp and the Docker image
public.ecr.aws/r4g1k2s3/vcon-dev/vcon-mcp. These pages describe release 1.9.2.
- What is the vCon MCP Server?: what it does, how MCP works, when to use it, how it relates to the conserver.
- Tool Reference: every tool with its key parameters, plus resources and prompts.
- Contract Tools: the envelope, paging, byte budgets and error codes of the LLM-facing tools.
- Transport and Deployment: transports, keys, tool profiles, tenancy, environment variables, Docker and npm.
- How the vCon MCP Server is Built: validation, plugins, the write and read paths, tables and search.
- Field-Name Migration: how
criticalandamendedreplacemust_supportandappended.
- Source on GitHub
- mcp.conserver.io, the generated technical reference