v0.4.0 — first public release

projectmind v0.4.0 — first public release

Persistent, compact project memory for AI coding agents. Your agent reads one
~400-token digest instead of re-scanning the codebase every session — and a
local ledger shows you exactly how many tokens (and dollars) that saved.

Highlights

• Measured savings, not promises — 78.9% (~1,541 tokens/session) on the
  reproducible benchmark (npm run benchmark); a gitignored local ledger
  records real savings as your agent works (projectmind savings), converted
  to dollars with zero config.
• Session handoff (unique to projectmind) — mind_handoff carries "what I
  was doing, what's next" across sessions, machines, and even different AI
  tools. It leads the next session's digest.
• One-command wiring — projectmind setup configures Claude Code, Cursor,
  Windsurf, Gemini CLI, Codex/AGENTS.md, and GitHub Copilot, idempotently.
• Offline, enforced — a CI test fails the build if any network API appears
  in the source. Two runtime dependencies. No telemetry, no LLM calls.
• Trust mechanics — atomic schema-validated writes, corruption self-heal
  with backups, drift detection (doctor), git-hook + watch-mode freshness.
• Cross-platform — CI on Linux, macOS, Windows × Node 18/20/22.
• VS Code companion — a status-bar counter showing your savings live
  (integrations/vscode, .vsix attached to this release).

Install

npm install -g @nodemint/projectmind
cd your-repo
projectmind init --seed && projectmind setup

Full docs, benchmark methodology, and the trust FAQ: see the README.