PAPI

PAPI (Persistent Adaptive Planning Intelligence) is a planning and memory layer for AI-assisted software development. It runs as an MCP server inside the AI tool you already use — no new app to open — and gives it structured, persistent project state: a live board, cycle plans, Active Decisions with confidence levels, build handoffs, and strategy reviews. What it does: - Plans what's next and generates BUILD HANDOFFs with scope, acceptance criteria, and guardrails — reducing drift and rework - Tracks every decision and the why behind it, so settled choices aren't re-argued - Keeps project state across sessions, tools, and subscriptions — pick up exactly where you left off - Runs strategy reviews every 5 cycles to check direction and compound learnings into the next plan Adds zero inference cost — it makes no LLM calls of its own; the reasoning happens inside the subscription you already pay for. Works with any MCP-compatible tool: Lovable, Bolt, Replit, Cursor, Claude Code, Windsurf, and ChatGPT. Setup is one connector link (or one CLI command) and takes about a minute. Free in alpha.