SENTINEL — Surveillance Contract Intelligence Agent

Ask anything about U.S. government surveillance spending. Get answers in seconds.

🔴 Live Demo → sentinel.osintnet.uk

Google Cloud Rapid Agent Hackathon 2026 — MongoDB Track

---

What Is SENTINEL?

Every facial recognition and AI surveillance contract awarded by the U.S. federal government is a public record. None of them are easy to find. We found 249 of them — totaling $3.83 billion — and built an AI agent that lets anyone ask questions about government surveillance spending in plain English and get sourced, verified answers in seconds.

This is not a chatbot. SENTINEL is a reasoning agent that:

• Queries a real MongoDB database of 249 verified government contracts
• Uses multi-step reasoning to aggregate, filter, and analyze procurement data
• Surfaces relationships between vendors, agencies, and spending patterns
• Never hallucinates — every answer is grounded in real contract data

Built by Indica Independent Media, whose mission is to use information and knowledge to defend vulnerable communities against powerful entities.

---

Architecture

User Query
    │
    ▼
FastAPI (sentinel.osintnet.uk)
    │
    ▼
Google ADK 1.32 Runner
    │
    ├── Gemini 2.5 Pro (reasoning + natural language)
    │
    └── MongoDB MCP Server (mongodb-mcp-server)
            │
            ▼
        MongoDB Atlas
        ├── 249 federal contracts
        ├── $3.83B total value
        ├── 2dsphere geo index
        └── Vendor / agency aggregations

Tech Stack

Layer	Technology
AI Reasoning	Gemini 2.5 Pro via Google ADK 1.32
Agent Framework	Google Cloud Agent Development Kit (ADK)
MCP Partner	MongoDB MCP Server (mongodb-mcp-server)
Database	MongoDB Atlas M0 — sentinel cluster
Backend	FastAPI + Python 3.11 + Uvicorn
Frontend	Vanilla JS + Leaflet.js (dark map UI)
Infrastructure	Dell OptiPlex → Cloudflare Tunnel → CF Zero Trust
Domain	sentinel.osintnet.uk

---

The Data

SENTINEL's database contains 249 verified U.S. federal government contracts sourced from:

• SAM.gov (System for Award Management)
• USASpending.gov
• 18 months of OSINT research by Indica Independent Media

Highlights:

• 🔴 Palantir USG Inc: $1.93B in surveillance/AI contracts
• 🔴 Palantir Technologies: $1.71B additional contracts
• Department of Defense: $2.31B total surveillance spend
• 29 unique government agencies tracked
• All contracts geocoded by state with 2dsphere indexing

---

How The Agent Works

SENTINEL uses the MongoDB MCP Server as its primary tool for data access. When you ask a question, the ADK agent:

1. Plans — Gemini 2.5 Pro determines which MongoDB operations are needed
2. Executes — Issues find, aggregate, count commands via MCP
3. Reasons — Synthesizes results into a factual, sourced answer
4. Suggests — Proposes a follow-up query to dig deeper

Example Queries

"Who are the top 5 vendors by total contract value?"
"Which agencies spent the most on facial recognition?"
"Show me all Clearview AI contracts"
"How many contracts were awarded after 2022 over $10M?"
"What states have the most surveillance contractors?"

---

MongoDB MCP Integration

This project uses the official mongodb-mcp-server as the ADK agent's data superpowers:

from google.adk.agents import Agent
from google.adk.tools.mcp_tool import McpToolset
from google.adk.tools.mcp_tool.mcp_session_manager import StdioConnectionParams
from mcp import StdioServerParameters

root_agent = Agent(
    model="gemini-2.5-pro",
    name="sentinel_agent",
    instruction=SENTINEL_INSTRUCTION,
    tools=[
        McpToolset(
            connection_params=StdioConnectionParams(
                server_params=StdioServerParameters(
                    command="npx",
                    args=["-y", "mongodb-mcp-server", "--readOnly"],
                    env={"MDB_MCP_CONNECTION_STRING": os.environ["MONGO_URI"]},
                ),
                timeout=30,
            )
        )
    ],
)

MCP Tools Available to the Agent:

• find — Query contracts with complex filters
• aggregate — Group, sum, rank by any field
• count — Count matching records
• collection-schema — Schema introspection
• db-stats — Database overview

---

Local Setup

# Clone
git clone https://github.com/indicaindependent/sentinel.git
cd sentinel

# Python venv
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

# Node (for MongoDB MCP server)
node --version  # requires v18+

# Environment
cp .env.example .env
# Add your keys:
#   GEMINI_API_KEY=your_key_from_aistudio.google.com
#   MONGO_URI=your_mongodb_atlas_connection_string

# Run
uvicorn main:app --host 0.0.0.0 --port 8001 --reload

Open http://localhost:8001

---

API Reference

GET  /api/health       — System status + engine info
GET  /api/stats        — Contract totals, top vendors/agencies
GET  /api/contracts    — Query contracts (search, vendor, agency, min_value)
POST /api/query        — Natural language query to Gemini + MongoDB MCP agent

---

Judging Criteria Alignment

Criterion	SENTINEL's Answer
Technological Implementation	Gemini 2.5 Pro + ADK 1.32 + MongoDB MCP — full partner stack
Design	Dark OSINT aesthetic, Leaflet map, responsive, zero UI friction
Potential Impact	Transparency tool for $3.83B in surveillance spend — accessible to any citizen
Quality of Idea	Real data, real problem — not a toy demo

---

License

MIT License — see LICENSE

---

About

Built by Indica Independent Media — a collective of technologists and artists who use knowledge and information to defend vulnerable communities.

"Every surveillance contract is a public record. We just made them readable."