What is piia-engram?

patdolitse/piia-engram — explained in plain English

Analysis updated 2026-05-18

★ 87PythonAudience · developerComplexity · 2/5LicenseSetup · easy

Mindmap

mindmap
  root((piia-engram))
    What it does
      Persistent AI identity memory
      Shared across AI tools
      Local JSON and Markdown storage
    Tech stack
      Python
      Model Context Protocol
      Local file storage
    Use cases
      Consistent coding standards
      Cross project knowledge reuse
      Auto extracted playbooks
    Audience
      Developers using multiple AI tools
      Vibe coders
      System architects

mindmap root((piia-engram)) What it does Persistent AI identity memory Shared across AI tools Local JSON and Markdown storage Tech stack Python Model Context Protocol Local file storage Use cases Consistent coding standards Cross project knowledge reuse Auto extracted playbooks Audience Developers using multiple AI tools Vibe coders System architects

What do people build with it?

USE CASE 1

Keep your coding standards and preferences consistent across Claude Code, Cursor, and Codex without repeating yourself.

USE CASE 2

Give a new AI coding session instant access to lessons learned and past architecture decisions from earlier projects.

USE CASE 3

Export your stored identity as a Markdown card to paste into tools that do not support MCP, like ChatGPT.

USE CASE 4

Automatically extract lessons and draft playbooks from completed multi-step workflows like deployments.

What is it built with?

PythonMCPJSONMarkdown

How does it compare?

	patdolitse/piia-engram	amazon-science/cyber-zero	thealgorithms/scripts
Stars	87	87	88
Language	Python	Python	Python
Last pushed	—	—	2023-10-04
Maintenance	—	—	Dormant
Setup difficulty	easy	hard	easy
Complexity	2/5	4/5	1/5
Audience	developer	researcher	ops devops

Figures from each repo's GitHub metadata at analysis time.

How do you get it running?

Difficulty · easy Time to first run · 30min

Installs via pip, each AI tool must be configured to point at the local MCP server, but no cloud account or external service is required.

Use freely for any purpose, including commercial use, as long as you keep the copyright and license notice.

So what is it?

piia-engram is a personal memory system for people who use multiple AI coding tools. The problem it addresses is straightforward: every time you open a new chat in Claude Code, Cursor, Codex, or any similar tool, that tool starts fresh with no knowledge of who you are, how you work, or what you have learned in previous sessions. piia-engram provides a single store of that context on your own computer, which any compatible AI tool can read at the start of each conversation. The data lives in a folder called ~/.engram/ as plain JSON and Markdown files you can open and edit yourself. What gets stored is not session logs or task history, but identity-level information: your coding style preferences, the standards you hold your code to, architectural decisions you have made and why, lessons from past mistakes, and notes about specific domains or projects you work in. The README draws a clear distinction between this kind of identity memory and agent memory systems like Mem0 or Zep, which record what happened in a task rather than who is behind every task. The tool connects to AI tools through MCP, which stands for Model Context Protocol, a standard that lets AI applications pull in external context at the start of a conversation. Once you configure piia-engram as an MCP server, tools that support the protocol automatically receive your stored context. For tools that do not support MCP, such as ChatGPT or Gemini, the tool can export a formatted Markdown card you paste in manually. Beyond passive storage, piia-engram has several active features. When you start a new project, a function called get_knowledge_inheritance looks across everything you have stored and returns the most relevant lessons and past decisions for that project. A session insight extractor can read a conversation summary and pull out new lessons to add to storage without manual note-taking. The tool also watches for multi-step workflows you complete, such as publishing a package or deploying a project, and drafts structured playbooks so the same steps are available the next time. Installation is via pip. Configuration involves pointing each AI tool to the local MCP server. All data stays on your machine with no cloud component or account required. The project is licensed under Apache 2.0.

Copy-paste prompts

Prompt 1

Install piia-engram with pip and configure it as an MCP server for Claude Code.

Prompt 2

Use get_knowledge_inheritance to pull relevant lessons into a new project I am starting.

Prompt 3

Extract session insights from this conversation summary and store them with piia-engram.

Prompt 4

Explain the difference between piia-engram's identity memory and task-based agent memory tools like Mem0.

Frequently asked questions