Sigillion-AI

Sigillion — AI skill for Claude Code

AI community

Lossless, local, provable memory for AI agents: every Claude Code session kept on your own computer, word for word, compressed and sealed, after compaction and after the 30-day transcript deletion.

How to install Sigillion

This entry records only its repository, not the path inside it, so there is no exact command to give. Open Sigillion-AI/sigillion and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Sigillion does

Lossless, local, provable memory for AI agents: every Claude Code session kept on your own computer, word for word, compressed and sealed, after compaction and after the 30-day transcript deletion. Claude Code first, every model next, then legal and finance. The waitlist, the releases, the website. Steddy Nova SRL.

Alternatives in AI

Pixel Agents — Compressed Reference — VS Code extension with embedded React webview: pixel art office where AI agents (Claude Code terminals) are an 6.5k ★
Tokentab — A CLI that reads Claude Code, Codex, and Gemini CLI session logs and works out how much they cost, by model, p 226 ★
Hipocampus — Drop-in memory harness for AI agents — 3-tier memory, compaction tree, hybrid search 149 ★

README

Sigillion

**Remember everything. Read almost nothing. Prove all of it.**

[![PyPI](https://img.shields.io/pypi/v/sigillion?label=PyPI&color=3f6b5e)](https://pypi.org/project/sigillion/) ![Python 3.10 to 3.14](https://img.shields.io/badge/Python-3.10%20to%203.14-23262b) ![Windows and Linux](https://img.shields.io/badge/Windows%20%7C%20Linux-x86__64-23262b) [![Website](https://img.shields.io/badge/website-sigillion--ai.github.io-7f5520)](https://sigillion-ai.github.io/sigillion/)

Sigillion is memory for AI agents, made by Steddy Nova SRL. It keeps the whole record of what an agent and a person did, word for word, on your own computer; compressed, so that a model can skim weeks of work in a few hundred tokens and open the full text only where it needs it; sealed, so that every record can be checked to show it was not altered. No telemetry: Sigillion itself sends nothing anywhere. What it hands Claude becomes part of your Claude Code session, which Claude Code sends to Anthropic as it does the rest.

One engine, in order

Claude Code first. Today Sigillion ships as a Claude Code plugin: every session is kept after compaction and after Claude Code deletes its transcripts, and Claude searches and reopens its own history while it works. Claude Code is where every feature is built, measured and proved on real work. It is the first step, not the whole idea.
Every AI model and agent tool next. The engine does not depend on one model. The same sealed memory will sit under other models and the tools people already use (coding agents, agent frameworks, assistants), with an SDK for anyone who builds agents.
**Then the fields where the record must stand: legal