A secure, enterprise-grade distributed surveillance and log ingestion architecture. The pipeline integrates a low-level Linux hardware implant, a symmetric cryptographic transit shield, a high-throughput asynchronous FastAPI data matrix server, and a Streamlit operational visualization panel.
- Kernel-Level Hooking: Intercepts character blocks directly from Linux subsystem files (
/dev/input/), bypassing application-isolation security wrappers on modern Wayland and X11 display servers. - Asynchronous Execution: Built on
asyncioandevdevevent loops to monitor input actions cleanly without causing CPU processing spikes or core hangs. - Background Exfiltration Tasks: Uses
asyncio.create_task()to schedule outbound HTTP transmissions in the background, ensuring target typing performance is never throttled during high network latency.
- Symmetric Cipher: Implements Fernet (AES-128 in CBC mode with an HMAC-SHA256 signature) via the
cryptographylibrary. - Integrity Enforcement: Guarantees secure data transport; payloads intercepted mid-transit trigger an instantaneous
InvalidTokenerror rather than rendering corrupt or manipulated strings.
- Asynchronous Routing: Powered by FastAPI to enable high-velocity concurrent ingestion pipelines across multiple reporting target nodes.
- Dependency Injection Security: Enforces custom perimeter security checks (
X-Argus-Tokenheaders) to completely drop invalid or unauthenticated scanning traffic. - Modern Lifespan Management: Employs SQLAlchemy 2.0 async context engines and
aiosqlitethread pools to separate disk operations from the web routing loops.
- Rapid-Response UI: Built with Streamlit and Pandas DataFrames to rapidly transform raw incoming JSON datastores into categorized, readable operational summaries.
- Isolated UI Refreshing: Features optimized
@st.fragmentrendering decorators to auto-pull updates every 3 seconds without causing global browser page flashing.
Since the agent communicates directly with system hardware lines, ensure Python headers and input tools are present:
sudo dnf install python3-devel kbd- Navigate to the project layout root directory:
cd ~/Projects/Colossus
- Build your local Python virtual environment configuration wrapper and activate it:
python3 -m venv .venv source .venv/bin/activate - Install the full engineering requirements sheet:
pip install fastapi uvicorn sqlalchemy aiosqlite cryptography httpx pydantic-settings streamlit pandas
Create a local, untracked environment variables tracking profile:
touch .envPopulate .env with your centralized secret variables mapping keys:
COLOSSUS_SHARED_KEY=PpGmRJE1b_16XobS-LVv2_mG7F8k7UendeG1exLxw3A=
ARGUS_API_TOKEN=super_secret_agent_password_123Launch the central broker engine:
uvicorn server:app --reload --port 8000API interactive swagger docs are dynamically accessible at http://127.0.0.
Fire up the real-time visual control dashboard in a separate tab session:
streamlit run dashboard.pyThe panel automatically launches on host browser port http://localhost:8501.
Execute the input listener agent on the target machine utilizing hardware access privileges:
sudo env "PATH=$PATH" python3 client.pyTo run this persistently in production environments, compile the agent using PyInstaller / PyArmor and configure it as a native background daemon system loop (/etc/systemd/system/argus.service).
When user input streams are committed and flushed over the network pipeline, the telemetry dashboard displays clean chronologically-tracked messaging cards:
🎯 Active Targets Monitored: 2 📊 Total Data Logs Intercepted: 6
🤖 Agent: linux-fedora
🕒 2026-08-25 21:10:42
Captured Keystrokes:
testing the secure api token hpf[BACKSPACE][BACKSPACE]nctionality; it works successfully.
This project architecture is constructed explicitly for authorized endpoint monitoring simulations, academic security engineering courses, or credentialed penetration testing frameworks. Unauthorized execution or tracking on computing equipment without explicit administrative ownership consent is unlawful and unethical.