> pmpt

Documentation

Everything you need to get started with pmpt

Installation

$ npm install -g pmpt-cli

Requires Node.js 18+. Includes both CLI and MCP server.

Recommended

Let your AI handle everything

With MCP connected, your AI automatically plans, saves, records decisions, and publishes — you just focus on building. The best pmpt experience.

  • AI auto-records every milestone with detailed descriptions
  • Planning, saving, publishing — all handled conversationally
  • Higher quality scores from richer development logs

Setup

pmpt mcp-setup

Auto-configure pmpt MCP server for your AI tool. Detects binary path and writes config — solves nvm PATH issues.

Usage: pmpt mcp-setup 
$ pmpt mcp-setup
# Detects pmpt-mcp path → select AI tool → done!
pmpt-mcp

Start pmpt MCP server for AI tool integration (Claude Code, Cursor, etc.). Included with pmpt — no separate installation needed.

Usage: pmpt-mcp 
# pmpt-mcp is included when you install pmpt:
npm install -g pmpt-cli

# Add to .mcp.json:
{
  "mcpServers": {
    "pmpt": { "command": "pmpt-mcp" }
  }
}

MCP Tools

pmpt_plan_questions

Get planning questions to ask the user conversationally

Usage: pmpt_plan_questions
Options:
  • projectPath — Project root path (defaults to cwd)
# AI gets questions, asks user one by one
pmpt_plan

Submit collected answers to generate AI prompt and project docs

Usage: pmpt_plan
Options:
  • projectPath — Project root path (defaults to cwd)
  • projectName — Project name
  • productIdea — Core product idea
  • coreFeatures — Key features (comma-separated)
  • additionalContext — Extra context (optional)
  • techStack — Preferred tech stack (optional)
# AI submits answers → plan.md, pmpt.md, pmpt.ai.md generated
pmpt_save

Save a snapshot with detailed summary after completing features, fixes, or milestones

Usage: pmpt_save
Options:
  • projectPath — Project root path (defaults to cwd)
  • summary — Detailed description of what was accomplished
# AI calls this automatically after completing work
pmpt_update_doc

Check off features, add progress notes, or backfill missing version summaries

Usage: pmpt_update_doc
Options:
  • projectPath — Project root path (defaults to cwd)
  • completedFeatures — Feature names to check off
  • progressNote — Detailed note to add to Snapshot Log
  • snapshotVersion — Version label (auto-generated if omitted)
# AI marks "User login" as done after implementing it
pmpt_log_decision

Record an architectural or technical decision with reasoning

Usage: pmpt_log_decision
Options:
  • projectPath — Project root path (defaults to cwd)
  • title — Decision title
  • reasoning — Why this decision was made
# AI logs "Chose SQLite — serverless environment"
pmpt_read_context

Read full project context (plan, docs, history, quality) to resume work

Usage: pmpt_read_context
Options:
  • projectPath — Project root path (defaults to cwd)
  • includeDocsContent — Include full docs content (default: true)
# AI reads context at session start to pick up where it left off
pmpt_status

Check tracked files, snapshot count, and quality score

Usage: pmpt_status
Options:
  • projectPath — Project root path (defaults to cwd)
# AI checks project status before starting work
pmpt_history

View version history with git commit info

Usage: pmpt_history
Options:
  • projectPath — Project root path (defaults to cwd)
  • limit — Max snapshots to return (most recent)
# AI reviews what changed in recent versions
pmpt_diff

Compare two versions, or a version against working copy

Usage: pmpt_diff
Options:
  • projectPath — Project root path (defaults to cwd)
  • v1 — First version number
  • v2 — Second version (omit to compare against working copy)
# AI compares v1 to current working copy
pmpt_edit_plan

Edit existing plan fields (productIdea, coreFeatures, techStack, etc.). Regenerates plan.md and pmpt.ai.md while preserving pmpt.md progress

Usage: pmpt_edit_plan
Options:
  • projectPath — Project root path (defaults to cwd)
  • projectName — New project name (optional)
  • productIdea — New product idea (optional)
  • coreFeatures — New core features (optional)
  • techStack — New tech stack (optional)
  • additionalContext — New additional context (optional)
# Translate productIdea to English:
pmpt_edit_plan(productIdea: "A Chrome extension...")
pmpt_quality

Check quality score and publish readiness

Usage: pmpt_quality
Options:
  • projectPath — Project root path (defaults to cwd)
# AI checks if project is ready to publish
pmpt_publish

Publish project to pmptwiki.com (requires pmpt login once)

Usage: pmpt_publish
Options:
  • projectPath — Project root path (defaults to cwd)
  • slug — Project slug
  • description — Project description (optional)
  • tags — Tags array (optional)
  • category — Project category (optional)
  • productUrl — Product URL (optional)
  • productUrlType — git or url (optional)
# AI publishes finished project to pmptwiki

MCP Workflow

$ pmpt mcp-setup # One-time setup
# Start your AI tool — it handles everything:
# plan → build → save → publish
$ pmpt login # Authenticate once for publishing