by CodeAbra
A cyber brain for your AI. It never forgets a detail, remembers exactly what you said, and learns how you work over time. Free, local, works with Cursor, Claude Code, Codex, OpenClaw, Hermes and more. MIT.
# Add to your Claude Code skills
git clone https://github.com/CodeAbra/iai-personal-memory-engineGuides for using ai agents skills like iai-personal-memory-engine.
Last scanned: 6/4/2026
{
"issues": [],
"status": "PASSED",
"scannedAt": "2026-06-04T08:21:17.000Z",
"npmAuditRan": true,
"pipAuditRan": true
}iai-personal-memory-engine is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by CodeAbra. A cyber brain for your AI. It never forgets a detail, remembers exactly what you said, and learns how you work over time. Free, local, works with Cursor, Claude Code, Codex, OpenClaw, Hermes and more. MIT. It has 834 GitHub stars.
Yes. iai-personal-memory-engine passed SkillsLLM's automated security scan — a dependency vulnerability audit plus prompt-injection heuristics — with no high-severity issues. You can read the full report in the Security Report section on this page.
Clone the repository with "git clone https://github.com/CodeAbra/iai-personal-memory-engine" and add it to your Claude Code skills directory (see the Installation section above).
iai-personal-memory-engine is primarily written in Python. It is open-source under CodeAbra on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh iai-personal-memory-engine against similar tools.
No comments yet. Be the first to share your thoughts!
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM is an informational directory and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
English | 中文
Your AI assistant forgets you every session. iai-pme gives it a memory that doesn't.
The memory style is autistic — by design: verbatim over paraphrase, precise cues, rare events kept rare. (Why the name)
Independent Autistic Intelligence — a personal memory engine. Fully local, ambient. Works with Claude Code, Claude Desktop, Cursor, Codex CLI, Gemini CLI, Cline, Continue.dev, Zed, Cherry Studio, Goose, Aider, Hermes, OpenClaw, Le Chat, Kimi — anything that speaks MCP-over-stdio.
A local server that speaks the MCP protocol and gives Claude, and any other MCP-compatible assistant, a long-term memory. It captures every turn of every session verbatim, organizes those captures over time into a personal map of who you are, and serves a small slice of relevant memory back at the start of each new conversation. You never have to say "remember this" or "what did we say last time?".
I built this for myself. It worked. I've been running it daily for months, and now I'm sharing it. The benchmarks were mostly for my own curiosity. I wanted to know if it actually works or if I'd just gotten used to it.
Under the hood it's not a wrapper around someone else's vector store and graph library — the parts that matter are my own code: the storage engine, the community-detection algorithm, the hyperdimensional memory substrate, and a native engine that makes it fast. More on that in Built our own.
And unlike cloud memory services, there's no API key, no account, and no telemetry: the engine, the store, and the embeddings all run locally. The only thing that leaves your machine is the normal model call your CLI already makes.
It also pays for itself in tokens. Memory injected at session start is context your assistant doesn't have to rebuild — no re-reading files it already read yesterday, no asking you the same orientation questions. In effect the engine doubles as a local context provider for your MCP host, and the dashboard keeps honest count: memory packs served, tokens injected, and a lower-bound estimate of tokens saved. For scale, measured on the author's own store over three recent weeks: 282 memory packs served (≈99,000 tokens of context injected) displaced agent searches that average 2,639 tokens a call — a lower-bound saving of ≈707,000 tokens, by the engine's own conservative formula. Per pack served that makes memory ≈88% cheaper (~350-token pack vs a ~2,850-token search round-trip, which is the measured 2,639 plus the per-call overhead). Your dashboard keeps your own count, live, and it measures the search side rather than assuming it.
That figure is about the packs the engine injects on its own — session start and the per-turn foresight pack. An explicit memory_recall you call mid-session is a different object: it is bounded by budget_tokens (default 1,500) and typically returns more than an ambient pack, so it saves real tokens against a search but not 88% of them.
| 🟢 Just want it to work? | 🔵 Want the numbers? | 🟣 Want the internals? |
| Install once, then forget it's there — no commands, fully local. | Every claim ships with the harness that proves it — run them yourself. | We built our own storage engine, clustering, HD substrate and Rust core. |
| → Quick start | → Benchmarks | → Built our own |
Mem0, Supermemory, Graphiti and Letta are memory layers for products you build — good at what they do, cloud-first, LLM-extraction pipelines. claude-mem is compression-based — it stores AI-generated summaries of your sessions (its own description). iai-pme is neither: it's a personal memory engine for the assistant you already use.
| iai-pme | typical memory layer | |
|---|---|---|
| What gets stored | Every turn, verbatim — write-once, never rewritten | LLM-extracted facts or compressed summaries |
| When a fact changes | Old version archived and still retrievable — Rescue@10 1.000, historical wording 1.000 | Profile updated; history retention varies |
| Where it runs | Your machine only — no account, no API key, embeddings computed locally | Managed cloud, or OSS + external LLM and vector DB |
| Storage | Own Rust engine — nothing external to install | Qdrant / Neo4j / Postgres / managed store |
| Cost of a memory write | File IO + a local embedding — no LLM in the capture path | An LLM call per extraction |
| Price | MIT. Everything is in this repo | OSS core + paid managed tier |
If you need multi-tenant memory for an app you're shipping, use one of them — honestly. If you want your assistant to remember you, that's this repo.
One command wires the whole thing — engine, capture hooks, MCP server, background daemon. macOS or Linux:
curl -fsSL https://raw.githubusercontent.com/CodeAbra/iai-personal-memory-engine/main/scripts/bootstrap.sh | bash
It checks your prerequisites, clones the repo to ~/.local/share/iai-pme, builds, installs the background engine and the capture hooks, registers the MCP server with Claude Code, and finishes with a health check. Re-run it any time to update. Prefer to look before you leap: add --dry-run to print every step without changing anything, or --preflight-only to check just the prerequisites.
Claude Code users can go through the plugin instead — it wires the MCP server and ambient capture in two lines:
/plugin marketplace add CodeAbra/iai-personal-memory-engine
/plugin install iai-memory@iai-pme
Restart the session and work normally: capture and recall are automatic from there. (pip install iai-pme first — the plugin carries the wiring, the package carries the engine.)
From PyPI — a prebuilt wheel carries the engine, so no clone and no toolchains:
pip install iai-pme
iai-mcp crypto init # one-time: creates the encryption key
That gives you the engine, the iai CLI, the dashboard and the MCP serv