PyP6Xer MCP Server

AI-agent tools for Primavera P6 XER schedules. Load, analyse, compare, edit, and export .xer files from Claude, ChatGPT, Cursor, or any MCP-compatible client.

PyP6Xer MCP is workflow-oriented, not just a parser. Unlike basic XER readers, it gives AI agents the full project-controls loop: schedule health checks, critical path, delay comparison, progress updates, relationship edits, and write-back to XER.

Prefer a web interface? p6.bouch.dev — upload an XER and analyse it without configuring MCP.

Use cases

• Ask questions about a Primavera P6 XER schedule
• Run critical path and float analysis
• Compare baseline vs update XER files and identify slipping activities
• Run schedule quality and health checks
• Edit activity fields safely and batch-update progress
• Export modified XER files
• Analyse relationship changes between two XER snapshots
• Build AI workflows for delay analysis and project controls

Example prompts

• Load this XER and show the critical path.
• Compare baseline.xer and update-03.xer and list major slippages.
• Find activities with high float, missing logic, or long durations.
• Update activity A1020 to 60% complete and export the edited XER.
• Analyse relationship changes between two XER snapshots.

Connect

Hosted (no install)

{
  "mcpServers": {
    "pyp6xer": {
      "type": "http",
      "url": "https://pyp6xer-mcp.fly.dev/mcp"
    }
  }
}

Local (uvx — no clone needed)

{
  "mcpServers": {
    "pyp6xer": {
      "type": "stdio",
      "command": "uvx",
      "args": ["pyp6xer-mcp"]
    }
  }
}

Local (from source)

Clone the repo, then point your MCP client at it:

git clone https://github.com/paulieb89/pyp6xer-mcp.git
cd pyp6xer-mcp
uv sync

{
  "mcpServers": {
    "pyp6xer": {
      "type": "stdio",
      "command": "uv",
      "args": ["run", "server.py"],
      "cwd": "/path/to/pyp6xer-mcp"
    }
  }
}

Usage

1. pyp6xer_load_file       — load a .xer from a local path, URL, or base64
2. <analysis tools>        — analyse the loaded schedule
3. pyp6xer_write_file      — persist any edits back to disk

Multiple files can be loaded simultaneously using different cache_key values.

Tools (29)

... View full README on GitHub