ClawSoulsClawSouls
SoulClaw Logo

SoulClaw

Soul-Powered AI Agents

Four products. One platform. Give your AI agent a persistent personality, shared memory, and security scanning.

⌨️

SoulClaw CLI

For terminal users & automation

Full-featured AI agent in your terminal. 4-Tier Soul Memory architecture (Soul → Core → Working → Session) with DAG conversation store and automatic promotion detection. Tiered Bootstrap loads only what’s needed — up to 60% fewer tokens per session. Semantic memory search powered by Ollama bge-m3.

Soul Memory 4-Tier (Soul → Core → Working → Session)
Semantic Memory Search (Ollama bge-m3)
DAG Conversation Store — lossless memory
Promotion Detector — auto T2→T1 promotion
Tiered Bootstrap — up to 60% fewer tokens
Gateway architecture + channel adapters
Telegram / Discord / Slack / Signal
Session Lifecycle Hooks (auto memory save)
Multi-agent sessions + sub-agents
SoulScan security scanning
$ npm install -g soulclaw
NEW
🧩

SoulClaw for VSCode

For developers & IDE users

Zero-setup AI agent inside your editor. Embedded engine calls LLM APIs directly — no gateway needed. Visual UI for Swarm Memory, Soul Checkpoints, and SoulScan. Team knowledge sharing with Git-based memory branches.

Swarm Memory (Git branch + LLM merge)
Soul Checkpoints (save/restore snapshots)
Visual SoulScan panel with inline warnings
Embedded engine — direct API calls
Age encryption for team memory
Model selection (Claude, GPT, Ollama)
ext install clawsouls.soulclaw-vscode
Available Now
📱

SoulClaw Mobile

For Android users

AI agent in your pocket. 80+ soul personas with persistent memory. On-device LLM support via llama.cpp. Chat anywhere with your personalized AI companion.

80+ Soul personas from marketplace
On-device LLM (llama.cpp + Vulkan)
Cloud LLM support (Claude, GPT, etc.)
Persistent memory across sessions
SoulScan security scanning
Coming Soon
☁️

ClawSouls Hosting

For everyone — no setup needed

Your AI agent, running 24/7 in the cloud. No terminal, no code, no server. Just pick a soul, connect your messenger, and go.

24/7 cloud hosting
Web chat + Telegram/Discord/Slack
Persistent memory across sessions
80+ Soul personas from marketplace
BYOK — bring your own LLM API key
7-day free trial
$7/mo per agent
Coming Soon

Compare

⌨️ SoulClaw CLI🧩 SoulClaw for VSCode
Killer Feature4-Tier Soul Memory + DAG StoreSwarm Memory + Soul Checkpoint + Visual UI
LLM EngineGateway-based, channel adaptersEmbedded engine, direct API calls
Target UsersTerminal users, server/automationIDE users, developers
Token EfficiencyTiered Bootstrap — up to 60% fewer tokensToken-based compaction
ChannelsTelegram, Discord, Slack, Signal, iMessageVSCode chat panel
Memory4-Tier + DAG + Semantic search + PromotionGit branches + LLM merge + encryption
SoulScanCLI scan + cron + web UISidebar panel + inline diagnostics
Team CollabMulti-channel, multi-sessionSwarm branches + age encryption
Setupnpm install + configInstall extension + API key
PlatformmacOS, Linux, Windows (WSL)Any OS with VSCode

One Platform, Shared Ecosystem

Both products share the same Soul Spec, the same marketplace, and the same security scanning engine.

👤

Soul Spec v0.5

Open standard for AI agent personality. Works with any LLM.

🏪

Soul Marketplace

Browse, install, and share agent personas. Community-driven.

🔍

SoulScan™

58 security rules. PII detection, prompt injection, identity drift.

Free & Open Source

Community edition is free forever. Enterprise features coming soon.

SoulClaw CLI is built on OpenClaw (MIT).