New in 0.3 · Pi Core support
Packwright compiles one agent definition — rules, memory, skills, workspace — into native packs for Pi, Codex, Claude Code, and Cursor. When you migrate, its memory, workspace, knowledge, and sources move with it.
Local-first. Zero telemetry. Packwright itself makes no network requests. Your coding runtime's data policy still applies.
Start with what you need
Packwright covers three common starting points. Customize responsibilities, capabilities, voice, boundaries, and emotional feedback—then answer a few simple questions to create your agent.
Expert engineer: Builds, reviews, debugs, tests, and ships technical work.
Versatile assistant: Plans projects, drafts deliverables, clarifies decisions, and keeps execution moving.
Personal secretary: Supports daily routines, life decisions, travel planning, and emotional support.
Need something more specific? Describe your own agent, or use packwright adopt to inventory and adapt an existing agent for migration.
Every AI coding tool wants its own instruction format. Packwright compiles one agent definition into native packs instead:
One agent definition projects into a native pack for each runtime.
score grades structure; doctor reports runtime readiness separately.
doctor --fix repairs drift without touching memory.
Terminal person? Drive it by hand. Not? Hand one prompt to your coding agent and just talk.
memory/, workspace/, knowledge/, and sources/. Excluded: CLAUDE.md and Claude Code-only files. Proceed?project/nova-codex · structure score 100.0. Runtime readiness is reported separately.Paste one prompt into Pi, Codex, Claude Code, or Cursor: USE_WITH_YOUR_AGENT.md
Eight public commands, plain files, no daemon. Quickstart ↓
Demo path · Claude Code → Codex
migrate --dry-run reports what the destination runtime can't carry — without writing anything.
packwright migrate project/nova-claude --to codex --target project/nova-codex --dry-runSee the full carry/exclude report. Nothing is written.
packwright migrate project/nova-claude --to codex --target project/nova-codex --yesmemory/, workspace/, knowledge/, and sources/ carry with hash verification.
packwright doctor project/nova-codex
packwright score project/nova-codexStructure and runtime-readiness evidence are reported separately. Receipts included.
Install the current release, choose a preset and name, then build a native target.
python -m pip install packwright==0.3.1 packwright init --template code --name Nova -o work/nova packwright build work/nova --adapter pi -o pack/nova-pi packwright install pack/nova-pi --adapter pi --target project/nova-pi