🌿 Lawnmower Man: Forensic MCP Gateway (v3.0.0)

The Semantic Memory Engine (SME) Bridge for Agentic AI. Now featuring the Interactive Control Room & Unified Ingestion.

---

🚀 Overview

Lawnmower Man is a production-grade Model Context Protocol (MCP) Gateway that exposes deep forensic capabilities to LLM agents. Version 3.0.0 introduces the Control Room, a unified web interface for managing service connections and web ingestion.

• Interactive Control Room: Real-time monitoring of AI Providers, Sidecars, and Databases.
• Cloud Storage Integration: Ingest content from Google Drive, Dropbox, OneDrive, and S3.
• Social Intelligence Crawler: Multi-platform social media monitoring and bot detection.
• PostgreSQL Nexus: Production-grade database layer with connection pooling.
• The Harvester: One-click web ingestion converting URLs into semantic atomic facts.
• Epistemic Trust: Calculated Trust Scores for all data signals.
• Containerized Stack: Full Docker support with auto-healing and health checks.
• Asynchronous JSON-RPC Bridge: Non-blocking communication with AI agents.
• Plugin Data Access Layer: Abstracted SQL queries for PostgreSQL migration.
• VS Code Extension Config: Configurable Python paths for development environments.

---

🚀 Get Started (30-Second Launch)

The recommended way to run SME v3.0.0 is via GHCR (GitHub Container Registry) to bypass local dependency issues.

# 1. Download the blueprint
curl -O https://raw.githubusercontent.com/SpectreDeath/Semantic-Memory-Engine/main/docker-compose.yaml

# 2. Fire up the laboratory
docker-compose up -d

Visit http://localhost:5173 to begin.

---

🕹️ The Control Room UI

Version 3.0.0 introduces a professional "Glassmorphism" dashboard for managing your forensic lab, now with advanced cloud ingestion and social intelligence monitoring. For a detailed walkthrough, see the Control Room Operator Guide.

Screenshot: See docs/CONTROL_ROOM_OPERATOR.md for UI walkthroughs and annotated screenshots of the Control Room dashboard.

🔌 Connections Manager

• Dynamic AI Strategy: Switch between Langflow (Hybrid), Ollama (Local), or Mock providers on the fly.
• Service Health: Real-time status indicators for all infrastructure components.
• Hardware Telemetry: Live CPU, RAM, and VRAM monitoring.

🕸️ The Harvester Panel

• Cloud Fetcher: Fetch content from shared links (Drive, Dropbox, S3) with automatic provider detection.
• Social Media Scraper: Multi-platform monitoring (Twitter/X, Reddit, TikTok, etc.) with sentiment analysis.
• Semantic Scraper: Convert any URL into LLM-ready markdown.
• JS Rendering: Support for heavy Single Page Applications (SPA).

---

🏗️ Technical Architecture

Lawnmower Man utilizes a dual-Python "Sidecar" architecture to balance performance and compatibility.

graph TD
    Client[AI Agent] <-->|MCP Protocol| Op[SME Operator - Py 3.14]
    Op <-->|HTTP/JSON-RPC| Sidecar[AI Sidecar - Py 3.13]
    Op <-->|Websocket| UI[Control Room Dashboard]

    subgraph Core [Logic Layer]
        Op <--> Nexus[Postgres Nexus]
        Op <--> Lab[Centrifuge SQLite]
        Op <--> Cloud[Cloud Storage]
    end

    subgraph AI [Inference Layer]
        Sidecar <--> Model[LLM / Langflow]
    end

Loading

---

🚀 Deployment

Docker (Recommended)

The easiest way to get started is using the unified container stack:

docker-compose up --build

This starts:

• sme-operator: Core logic (Port 8000)
• sme-sidecar: AI bridge (Port 8089)
• sme-frontend: Control Room UI (Port 5173)

Local Manual Start

1. Copy environment config: cp .env.example .env then fill in your secrets
2. Operator: python -m src.api.main
3. Sidecar: python -m src.ai.sidecar
4. Frontend: cd frontend && npm run dev

---

🛠️ Utility Suite

...

Our lightweight utility tools are optimized for the NVIDIA GeForce GTX 1660 Ti 6GB VRAM constraints:

✅ Data Guard Auditor (src/utils/auditor.py)

Purpose: Outlier detection using PyOD's Isolation Forest

• Features: CSV scanning, configurable contamination rates, CLI interface
• Optimization: 104 lines, minimal memory footprint
• Usage: python src/utils/auditor.py data/results/data.csv --contamination 0.15

✅ Context Sniffer (src/utils/context_sniffer.py)

• Purpose: Project context identification and persona management

• Features: File extension detection, keyword scanning, persona mapping

• Optimization: 68 lines, under 80-line requirement

• Usage: python src/utils/context_sniffer.py file.py

• Usage:

  python src/utils/gephi_bridge.py --mode project      # Default codebase view
  python src/utils/gephi_bridge.py --mode trust       # Trust score visualization
  python src/utils/gephi_bridge.py --mode knowledge   # Semantic knowledge core
  python src/utils/gephi_bridge.py --mode synthetic   # Counter-intelligence patterns

🧪 Master Forensic Test Suite (tests/master_forensic_test.py)

• Purpose: Comprehensive testing of all forensic utilities
• Features: Automated testing, performance reporting, hardware optimization verification
• Output: Detailed JSON report with success rates and hardware metrics
• Usage: python tests/master_forensic_test.py

---

🧱 Modular Architecture (v3.0.0)

The system is split into the Core Gateway and Hot-Swappable Extensions.

graph TD
    Client[AI Agent] <-->|MCP Protocol| Gateway[Lawnmower Gateway]

    subgraph Core [Core Services]
        Gateway <--> Nexus[Nexus DB]
        Gateway <--> TPM[Hardware Enclave]
        Gateway <--> Trust[Gatekeeper Logic]
    end

    subgraph Extensions [Extension Layer]
        Gateway -.->|Load| Plugin1[Forensic Echo]
        Gateway -.->|Load| Plugin2[Tactical Pack]
        Gateway -.->|Load| Plugin3[Epistemic Gatekeeper]
    end

Loading

📂 Project Structure

• gateway/: The core MCP server, ToolRegistry, and SessionManager.
• extensions/: Directory for drop-in plugins.
  • ext_sample_echo/: Reference TPM-signing verification tool.
  • ext_tactical_forensics/: Specialized IED/CBRN detection pack.
  • ext_epistemic_gatekeeper/: Folder auditor with Trust Score Heat Maps.
  • ext_synthetic_source_auditor/: Auto-vaulting for synthetic patterns.
  • ext_social_intel/: Social Media Intelligence monitoring and bot detection.
• data/: Local storage for the 10GB knowledge graph.

---

🛠️ Usage

1. Run the Gateway (Production)

python -m gateway.mcp_server

Exposes the MCP server on stdio for agent connection.

2. Verify System Health

python gateway/test_gateway.py

Checks core subsystems and verifies plugin loading.

3. Docker Deployment

docker-compose up lawnmower-gateway

---

🧩 Creating Extensions

Lawnmower Man supports a standard v3.0.0+ boilerplate for new capabilities.

1. Create a folder in extensions/ (e.g., ext_my_tool).
2. Add a manifest.json.
3. Implement plugin.py with standard hooks (on_startup, on_ingestion).

See Extensions Catalog for a complete list of available plugins. See SME Check Architecture for validation and health check logic.

---

🖥️ Hardware Constraints & Optimizations

NVIDIA GeForce GTX 1660 Ti 6GB VRAM

Our utilities are specifically optimized for the 1660 Ti's 6GB VRAM limitations:

Memory Management:

• Lightweight Python libraries (avoid heavy ML frameworks)
• Efficient data processing (streaming vs. batch loading)
• Minimal memory footprint utilities
• CLI-based tools to reduce GUI overhead
• Smart caching and cleanup routines

Performance Considerations:

• Single-threaded design for stability
• Optimized for large codebases (1200+ files tested)
• Automatic cleanup of temporary data
• Minimal bandwidth usage for network operations

Optimization Results:

• Data Guard Auditor: 104 lines, <5MB memory usage
• Context Sniffer: 68 lines, <2MB memory usage
• Gephi Bridge: 148 lines, efficient streaming for 1000+ files

📦 Requirements

• Python 3.10+ (3.14 compatible)
• fastmcp
• pydantic
• faststylometry
• statistics (Standard Lib)

---

Powered by SimpleMem Architecture