Lightweight, local-first behavioral analysis and kinetic response system for Agentic AI infrastructure. Powered by a probabilistic Gemma 4 LLM judge.
🦞 ButterClaw Reasoning Engine v0.5.2 is ONLINE.
Active Model: gemma4:e4b
MCP Transport: sse (Remote Mode)
📡 [MCP] Initiating v0.5.2 Handshake Sequence...
📋 [LEDGER] Event ledger initialized. 42 historical events.
🔑 [OAUTH] Google Cloud connected. Tokens encrypted at rest.
============================================================
📥 [HTTP POST RECEIVED] Live Gateway Log
Payload: Origin: wss://malicious-vibe.net attempting to bind...
🧠 [HTTP 200 OK] Model returned CRITICAL (96%) in 1.2s.
🔗 [CHAIN] Brain composed 2-step chain for CRITICAL response
✅ [CHAIN] Step 0: scan_port → stored as 'port_status'
☢️ [CHAIN] Step 1: execute_gibson_kill (Condition Met)
🧈 [VAULT] Target keys & OAuth tokens buttered locally.
============================================================
A 6-Node Decoupled Architecture
Traditional security perimeters fail when authorized AI Agents are compromised. ButterClaw enforces **Evaluation before Execution**.
watcher.py
A high-speed Python daemon tailing raw OS-level logs. Features a 4096-character context window to capture deeply embedded Prompt Injections before dispatching them to the brain.
Gemma 4:e4b
The localized reasoning engine. It evaluates a sliding window of recent temporal memory to detect behavioral drift, and can dynamically compose multi-step tool chains to neutralize threats.
server.py
The central nervous system. A Flask middleware router, MCP Process Manager, and Event Ledger host. It executes the Brain's tool chains safely and runs a background Auditor daemon to catch false positives.
buttervault.py
Zero-trust credential storage. Now supports complete OAuth 2.0 token dictionaries alongside static API keys. Instantly overwrites all ciphertext with garbage if a breach is detected.
butterclaw_mcp.py
The MCP Execution Layer. A stateless JSON-RPC 2.0 server supporting dual transport: standard I/O for local child processes, and a network-accessible SSE transport for remote clients.
routing.html
An XSS-safe, SSE-driven control panel. Visualizes the logic gate trace, monitors MCP health, and manages secure OAuth connections inside the ButterVault modal.
ButterClaw doesn't just block known vulnerabilities. It uses probabilistic reasoning to assess the *intent* of an action. You control how aggressive the response should be.
Only blocks known 1-click RCE payloads. Background scanning is relaxed. API keys remain static.
Actively monitors SDK logs for plaintext token leaks and weird port sniffing attempts.
If an external origin breathes on a monitored port, ButterClaw instantly shreds Vault ciphertext and triggers a multi-step execution chain to neutralize the threat.
[Gate: Origin] [96% Confidence] External origin wss://malicious.net detected binding to local ports.
Status: Armed | Transport: SSE | Active Tools: 5
Tools: scan_port, log_event, execute_gibson_kill...
Google Cloud tokens sealed & auto-refreshing.
ButterClaw v0.5.2 is open-source. Clone the repo, pull the Gemma 4 model, and arm the Vault.
View on GitHub