JSPM

  • ESM via JSPM
  • ES Module Entrypoint
  • Export Map
  • Keywords
  • License
  • Repository URL
  • TypeScript Types
  • README
  • Created
  • Published
  • Downloads 287
  • Score
    100M100P100Q118279F
  • License MIT

Knowledge layer for AI agents. Claude remembers you. Phren remembers your work.

Package Exports

    This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (@phren/cli) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.

    Readme

    phren

    Your agents forget everything. Phren doesn't.

    npm version license

    Phren is a git-backed knowledge layer that gives AI agents persistent memory across sessions, projects, and machines. Findings, decisions, and patterns are captured as markdown in a repo you control — no hosted service, no vendor lock-in.

    Install

    npx @phren/cli init

    That single command creates ~/.phren, wires up MCP, installs hooks, and gives your agents a memory they can actually keep. Re-running on a new machine with an existing remote picks up right where you left off.

    What phren tracks

    • Findings — bugs hit, patterns discovered, decisions and their reasoning. Tagged by type ([pattern], [decision], [pitfall], [observation]) with per-type decay rates
    • Fragments — named concepts (auth, build, React) that connect findings across projects. When you search for a topic, phren pulls in everything linked to that fragment
    • Tasks — work items that persist across sessions with priority, pinning, and GitHub issue linking
    • Sessions — conversation boundaries with summaries and checkpoints, so the next session picks up where this one left off
    • Skills — reusable slash commands you teach phren. Drop them in ~/.phren/global/skills/ and they work everywhere

    How it works

    • Surfaces relevant context on every prompt via hooks. Agents build on what they know instead of starting fresh
    • Trust scores decay over time — old findings lose confidence. Decisions never decay. Observations expire in 14 days
    • Syncs across machines through git push/pull. No coordination service
    • Works with Claude Code, Copilot, Cursor, and Codex — one memory, every agent
    • Shell and web UI for browsing, searching, and triaging (phren or phren web-ui)

    Quick start

    npx @phren/cli init          # set up phren (interactive walkthrough)

    Init detects your tools, registers MCP servers, and installs lifecycle hooks. After it finishes, open a prompt in any tracked project — phren is already injecting context.

    To add a project later, run phren add from that directory. To browse what phren knows, run phren to open the interactive shell.

    Learn more

    MIT License. Made by Ala Arab.