skills.rest
kingcharleslzy-ai
kingcharleslzy-aiCOMMUNITY
Explore all skills by this author →
GitHub

agent-soul

Compile append-only NDJSON events into canonical memory files via Git.

5|Updated Mar 10, 2026
Search Tags:#ai agents#shared memory#event sourcing#git-native#persistent identity#canonical compilation#cross-runtime synchronization
One-click install
npx skills add https://github.com/kingcharleslzy-ai/agent-soul --skill agent-soul
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-soul
Source: https://github.com/kingcharleslzy-ai/agent-soul/tree/main
Command: npx skills add https://github.com/kingcharleslzy-ai/agent-soul --skill agent-soul

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and assets (resource) components.

What problem does it solve?

This Skill solves the problem of AI agents forgetting and desynchronizing across sessions, devices, and different runtimes by providing a shared, durable memory and persona stored in a user-controlled Git repository.

Core Features & Use Cases

  • Persistent identity (L0 Soul): keeps one consistent assistant persona across runtimes using SOUL.md, IDENTITY.md, USER.md, and VOICE.md.
  • Durable shared memory (L1 Memory): compiles event-based facts into canonical/profile.md and canonical/stable-memory.md.
  • On-demand context (L2 Context): loads project, fuzzy, and provenance context only when needed to keep context lightweight.
  • Append-only event surface with expiry: writes NDJSON events to sources/<source> and supports valid_until for time-bound facts.
  • Cross-agent synchronization: makes new learnings available to other agents by compiling sources into canonical/* via scripts and CI.
  • Auditability and portability: avoids databases and memory APIs by using Git as the storage and distribution layer.

Quick Start

Use the agent-soul skill by cloning your template into a private repo, filling SOUL.md, IDENTITY.md, USER.md, and VOICE.md, then telling your AI runtime to follow JOIN.md and to write join events with scripts/add_event.py using a stable source id.

Frequently Asked Questions about agent-soul

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I maintain persistent identity and shared memory for AI agents across different runtimes?▼

You can maintain persistent AI agent memory by using a Git-native approach that stores persona and event-based facts in a user-controlled repository, compiling durable memory into canonical surfaces like profile.md and stable-memory.md for cross-runtime synchronization.

What is the best way to synchronize memory across multiple AI agent sessions without using a database?▼

The best way to synchronize memory without a database is using Git as the storage and distribution layer. Agents write append-only NDJSON events to sources, which are compiled into canonical files for cross-agent synchronization and auditability.

How do I set up event sourcing for AI agents to compile canonical memory?▼

To set up event sourcing for AI agents, clone a template into a private repo, fill SOUL.md and IDENTITY.md, then instruct your AI runtime to write append-only NDJSON events using a stable source id and compilation scripts.

Does Git-native AI memory support time-bound facts and temporal expiry?▼

Git-native AI memory supports time-bound facts through temporal expiry. The system honors valid_until fields within the append-only event surface, ensuring outdated context is expired during canonical compilation.

Can I onboard multiple agent bodies for the same user using a shared Git repository?▼

You can onboard multiple agent bodies for the same user by compiling sources into canonical files via scripts and CI. This makes new learnings available to other agents while keeping L2 context lightweight by loading it on demand.

When should I not use a Git-native approach for AI agent memory synchronization?▼

You should not use a Git-native approach for AI agent memory synchronization if your use case requires real-time database APIs or cannot tolerate Git repository overhead for high-frequency event writes and canonical compilation.

Related Skills

Explore More Software Engineering Skills→
addisonkaddisonk
1

openclaw-workspace-authoring

Author and review OpenClaw agent workspace files like SOUL.md and AGENTS.md.

Community
Advanced
Abhishekojha38Abhishekojha38
3

Memory

Store structured facts and preferences in MEMORY.md and log events to HISTORY.md.

Community
Intermediate
haJ1thaJ1t

pr-babysitter

Monitors open pull requests, fixes CI errors, replies to reviews, resolves conflicts, and ensures mergeability via GitHub API and git automation.

Community
Advanced
oimiragieooimiragieo
36

memory-manager

Synchronize memory patterns across CLAUDE.md and the Memory Tool.

Community
Advanced