PlanExe

PlanExe exposes an MCP server that lets AI agents interact with PlanExe to generate and retrieve strategic plans. Using the remote MCP path, you point your MCP client at the PlanExe endpoint and authenticate with your API key. The workflow supports initiating plan creation, polling for status, retrying failed plans, and downloading results. Typical tool usage follows the documented sequence: start with example_plans (optional preview), then use example_prompts to craft prompts, select or customize a model_profile with model_profiles if needed, draft and approve the prompt, call plan_create, poll with plan_status until completion, optionally retry with plan_retry, and finally download artifacts via plan_download or plan_file_info. This setup focuses on turning a plain-English goal into a structured, long-form plan with sections like executive summary, governance, risk registers, and more. Be aware that PlanExe outputs are starting points and often need refinement for budgets, timelines, and regulatory specifics.

To get started, obtain an account at PlanExe, acquire an API key (pex_... format), and configure your MCP client to use the remote endpoint. You can integrate this into automation or agent workflows, allowing agents to generate plans as part of larger product or project workflows. If you plan to run local or proxy-based downloads, you can switch to the mcp_local variant described in the documentation, but for most use cases the remote endpoint provides the quickest path to plan generation and retrieval.

Prerequisites:

• An API key from PlanExe (PEX). Ensure you have an active PlanExe account with sufficient funds to create plans.
• A compatible MCP client that can consume the MCP schema (any tool supporting PlanExe MCP responses).

Step 1: Retrieve your API key

• Log in to PlanExe and navigate to your account to obtain the API key (starts with pex_).

Step 2: Configure your MCP client

• Use the following MCP configuration to connect to PlanExe:

{
  "mcpServers": {
    "planexe": {
      "url": "https://mcp.planexe.org/mcp",
      "headers": {
        "X-API-Key": "pex_your_api_key_here"
      }
    }
  }
}

Step 3: Test a quick plan (example)

• With your MCP client configured, issue a simple prompt like:
  • example_plans (optional): preview an example output
  • plan_create: provide a plain-English goal and any constraints
• Poll with plan_status until the plan_id reports as finished, then download with plan_download or plan_file_info.

Step 4: Optional local proxy or Docker-based setups

• If you prefer local downloads or a self-hosted proxy, follow the PlanExe MCP docs to configure mcp_local with your own endpoint and path. The remote endpoint remains the simplest path for most users.

Prerequisites recap:

• Open internet access from the client to https://mcp.planexe.org/mcp
• A PlanExe API key (pex_...)
• A compatible MCP client or library that can handle JSON-based MCP requests and responses

Tips and common considerations:

• API key security: treat your API key as a secret; do not commit it to code repositories. Use environment variables or secret management when integrating into automation.
• Plan creation is asynchronous. Each plan_create call returns a plan_id; poll plan_status periodically (every few minutes as recommended by PlanExe) until completion.
• The remote MCP endpoint may have rate limits or quota considerations based on your PlanExe account. Plan accordingly in automation to avoid hitting limits.
• If you need artifacts stored locally, consider using the mcp_local proxy option and provide PLANEXE_PATH for downloads.
• Use plan_retry if a plan fails; you may need to adjust prompts or model profiles to improve outcomes.
• Ensure you have network access to the PlanExe endpoint and that any corporate firewall or proxy allows outbound HTTPS traffic to mcp.planexe.org.