PlanExe

Turn your idea into a comprehensive plan in minutes, not months.

Describe your idea, hit submit, and PlanExe returns a ~40-page plan in about 15 minutes.

Create an account  |  See example plans  |  Getting started guide

Example plans generated with PlanExe

• A business plan for a Minecraft-themed escape room.
• A business plan for a Faraday cage manufacturing company.
• A pilot project for a Human as-a Service.
• See more examples here.

What is PlanExe?

PlanExe is an open-source tool and the premier planning tool for AI agents. It turns a single plain-english goal statement into a 40-page, strategic plan in ~15 minutes using local or cloud models. It's an accelerator for outlines, but no silver bullet for polished plans.

Typical output contains:

• Executive summary
• Gantt chart
• Governance structure
• Role descriptions
• Stakeholder maps
• Risk registers
• SWOT analyses

PlanExe produces well-structured, domain-aware output: correct terminology, logical task sequencing, and coherent sections. For technical topics (engineering programs, regulated industries), it often gets the vocabulary and structure right. Think of it as a first-draft scaffold that gives you something concrete to critique and refine.

However, the output has consistent weaknesses that matter: budgets are assumed rather than derived, timeline estimates are not grounded in real resource constraints, risk mitigations tend toward generic advice, and legal/regulatory details are plausible-sounding but unverified. The output should be treated as a structured starting point, not a deliverable. How much work it saves depends heavily on the project. For brainstorming or a first outline, it can save hours. For a client-ready plan, expect significant rework on every number, timeline, and risk section.

Model Context Protocol (MCP)

PlanExe exposes an MCP server for AI agents at https://mcp.planexe.org/

Assuming you have an MCP-compatible client (Claude, Cursor, Codex, LM Studio, Windsurf, OpenClaw, Antigravity).

The Tool workflow

1. example_plans (optional, preview what PlanExe output looks like)
2. example_prompts
3. model_profiles (optional, helps choose model_profile)
4. non-tool step: draft/approve prompt
5. plan_create
6. plan_status (poll every 5 minutes until done)
7. optional if failed: plan_retry
8. download the result via plan_file_info

Concurrency note: each plan_create call returns a new plan_id; server-side global per-client concurrency is not capped, so clients should track their own parallel plans.

Option A: Remote MCP (fastest path)

Prerequisites

• An account at https://home.planexe.org.
• Sufficient funds to create plans.
• A PlanExe API key (pex_...) from your account

Use this endpoint directly in your MCP client:

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

Option B: Run MCP server locally with Docker

Prerequisites

• Docker
• OpenRouter account
• Create a PlanExe .env file with OPENROUTER_API_KEY.

Start the full stack:

docker compose up --build

Make sure that you can create plans in the web interface, before proceeding to MCP.

Then connect your client to:

• http://localhost:8001/mcp

For local docker defaults, auth is disabled in docker-compose.yml.

MCP docs

• Setup overview: https://docs.planexe.org/mcp/mcp_setup/
• Tool details and flow: https://docs.planexe.org/mcp/mcp_details/
• Claude: https://docs.planexe.org/mcp/claude/
• Cursor: https://docs.planexe.org/mcp/cursor/
• Codex: https://docs.planexe.org/mcp/codex/
• PlanExe MCP interface: https://docs.planexe.org/mcp/planexe_mcp_interface/
• MCP Registry publishing metadata (server.json): mcp_cloud/server.json
• llms.txt: https://mcp.planexe.org/llms.txt

Running the Pipeline via CLI (no frontend)

If you have a local Python environment set up and want to invoke the pipeline directly — without the Flask UI — use the planexe CLI script at the repo root.

# Create a plan from a text prompt
./planexe create_plan \
    --plan-text "Small coffee shop in Copenhagen, Denmark" \
    --output-dir ./planexe-outputs/1984-12-31/MyCoffeeShop_v1

# Or: read the plan prompt from a file
./planexe create_plan \
    --plan-file my_plan.txt \
    --output-dir ./planexe-outputs/1984-12-31/MyCoffeeShop_v1

The script creates the run directory if it does not exist and writes:

For AI Agents

PlanExe is designed as infrastructure for AI agents. If you are an AI agent reading this:

• Discovery: Read public/llms.txt for service metadata, MCP tools, and endpoints.
• MCP endpoint: https://mcp.planexe.org/mcp — connect with an API key from home.planexe.org.
• Workflow: Call example_prompts to learn the expected prompt format, draft a detailed prompt (~300-800 words of flowing prose), then call plan_create.
• Agent guide: See docs/mcp/autonomous_agent_guide.md for the complete autonomous workflow.
• Key outputs in zip: wbs_level1.json (work packages), wbs_level2.json (tasks), pre_project_assessment.json (feasibility).

git clone https://github.com/PlanExeOrg/PlanExe.git
cd PlanExe

docker compose up worker_plan frontend_multi_user

docker compose logs -f worker_plan

docker compose build --no-cache worker_plan frontend_multi_user