Skip to content

developer module architecture

Jump to bottom
MD MUFTHAKHERUL ISLAM MIRAZ edited this page Jun 24, 2026 · 2 revisions

🏛️ Module Architecture

Curious how Siyarix ticks under the hood? This document breaks down the internal architecture, detailing how modules interact and the design patterns I used while building this personal passion project.

🧩 Module Organization

Siyarix is organized into directories and standalone modules. Here's a map:

Module Responsibility
core/ 🧠 The orchestrator kernel: AgentCore, SwarmRouter, and CommandPipeline.
chat/ 💬 The interactive REPL UI, session management, and LLM engine.
providers/ ☁️ AI provider abstraction layer and API management.
parsers/ 🔍 Tool output parsers.
plugins/ 🔌 Dynamic plugin loader.
output/ 🎨 The rendering engine.
report/ 📊 Security report generation.
stealth.py 🥷 Basic stealth configurations for tools.
dlp.py 🛡️ Data masking helper.
permission_gate.py 🚧 Command access control.
credential_store.py 🔐 Local encrypted credential vault.
workflow.py 🔄 Workflow execution logic.

⚙️ Execution Engine

The Execution Engine operates in three modes, dispatched by the Task Planner:

Mode Planner Permission Level Autonomy
REGISTRY Template-based Full gate None
AUTONOMOUS LLM-driven Minimal Full
HYBRID (Default) Combined Full gate User confirmation

🌊 Execution Flow

  1. The Planner generates an ExecutionPlan.
  2. Each step passes through the PermissionGate.
  3. Steps execute using asyncio.gather().
  4. Output is parsed via the ParserRegistry.
  5. Transient errors are handled by the ProviderStateManager.

Tip

Worker Pool Throttling: Siyarix uses asyncio.Semaphore to limit concurrency and prevent resource exhaustion.

🗺️ Task Planners

Siyarix uses a dual-planner architecture:

  • Registry Planner: A deterministic, template-based planner using keyword matching.
  • Autonomous Planner: The LLM-driven planner that translates natural language into execution steps.

🚧 Permission Gate (permission_gate.py)

To help prevent accidents, Siyarix runs a review before execution:

  1. Syntax Gate: Checks structural limits.
  2. Danger Analysis: Scans against potentially dangerous commands. It returns ALLOW, DENY, or asks the user for REVIEW.

☁️ Provider Manager (providers/manager.py)

The ProviderManager handles integrations with various AI providers (OpenAI, Anthropic, Gemini, Ollama, etc.). It includes failover logic so you can keep working if your primary provider goes down.

🔐 Credential Store (credential_store.py)

To keep your API keys safer, the CredentialStore encrypts them locally using AES-256-GCM and the OS system keyring, avoiding plaintext configuration files.

🛡️ DLP Engine (dlp.py)

The DLP engine is a lightweight tool to help prevent accidentally leaking secrets (like AWS keys or local passwords) to cloud providers by masking them before sending prompts.

🧠 Knowledge Graph (knowledge_graph.py)

An in-memory directed graph of discovered nodes (Hosts, ports, vulnerabilities) that helps provide context for the report engine.

🐝 Swarm Architecture (core/swarm.py)

A simple multi-agent setup where tasks are broken down between specialized sub-agents (Recon, Exploit, Report).

📈 Learning System (learning_system.py)

A privacy-preserving feature that tries to remember past successful executions and command structures to improve future runs locally, without relying on external ML frameworks.

Note

👋 Welcome to Siyarix! This is a personal passion project built by a single developer. It's currently under active development and growing fast. Expect rough edges, but lots of love! ❤️

🗺️ Siyarix Documentation Map

Welcome to the Siyarix Documentation Map! This page serves as your master compass for navigating the extensive documentation we have built for the platform.

Whether you are a brand new user, a seasoned security operator, or a developer looking to contribute to the core engine, you can find exactly what you need here.

🧭 Quick Navigation

Not sure where to start? Pick the path that best describes you:

🌱 For New Users

Just getting started? We highly recommend following these guides in order:

  1. Installation Guide — Get Siyarix running on your machine.
  2. Onboarding Wizard — Let our interactive wizard help you set up your API keys and environment.
  3. Setup & Configuration — A deeper dive into customizing your setup.
  4. Your First Run — A gentle walkthrough of your very first Siyarix command.

🛡️ For Security Operators

Ready to put Siyarix to work? Dive into our operational guides:

💻 For Developers & Contributors

Looking under the hood or wanting to write some code? Start here:

📂 The Complete Documentation Tree

If you prefer to browse the raw structure, here is a complete layout of the docs/ folder:

docs/
├── 🚀 getting-started/       # Installation, onboarding, and configuration
│   ├── installation.md       # Multi-platform install (pip, brew, winget, docker)
│   ├── onboarding.md         # The interactive 11-step setup wizard
│   ├── setup.md              # Managing API keys, credentials, and settings
│   ├── first-run.md          # A walkthrough of your first session
│   ├── configuration.md      # A deep-dive into advanced settings
│   └── troubleshooting.md    # Common issues and how to fix them instantly
│
├── 📖 user/                  # Daily operations and workflows
│   ├── cli-commands.md       # Reference for 50+ CLI commands across 12 groups
│   ├── interactive-chat.md   # Mastering the AI REPL and 54+ slash commands
│   ├── security-workflows.md # Recon, vulnerability assessment, incident response
│   ├── cloud-scanning.md     # Multi-cloud security scanning (under development)
│   ├── compliance.md         # Framework mapping (SOC 2, NIST, GDPR, PCI-DSS)
│   ├── threat-intelligence.md# Integrations with OTX, NVD, and MITRE ATT&CK
│   ├── playbooks.md          # Building automated YAML-based IR playbooks
│   ├── workflow-files.md     # DAG workflow reference (programmatic API)
│   ├── reporting.md          # Multi-format report generation
│   ├── offline-registry.md   # Running without AI (Offline/Registry execution mode)
│   └── ai-workflows.md       # Advanced AI-driven autonomous operations
│
├── 💻 developer/             # Building, testing, and extending Siyarix
│   ├── codebase-overview.md  # Full module structure mapping
│   ├── contribution-guide.md # How to submit PRs and our coding standards
│   ├── module-architecture.md# Component design and responsibilities
│   ├── testing.md            # Writing tests (pytest), coverage, and CI/CD
│   └── building.md           # Packaging, distribution, and Docker builds
│
├── 🏗️ architecture/          # System design and core internals
│   ├── overview.md           # High-level data flow and layered orchestration
│   ├── ai-agent-pipeline.md  # The AgentCore reasoning and execution pipeline
│   ├── provider-abstraction.md# How we unify 26 different AI providers
│   ├── execution-engine.md   # Plan-based step orchestration
│   ├── memory-and-state.md   # Knowledge graph, session persistence, and learning
│   ├── security-model.md     # The Permission Gate, DLP, audit logging, and OPSEC
│   └── intent-routing.md     # Semantic intent classification and routing
│
├── 🧠 ai/                    # Deep dive into the AI provider & agent systems
│   ├── routing.md            # Managing 26 providers, failovers, and circuit breakers
│   ├── persona-system.md     # Overview of our 10 security personas
│   ├── agent-reasoning.md    # The Observe-Reason-Act loop and tool call repair
│   ├── tool-execution.md     # The tool registry, capability graph, and parsers
│   ├── ensemble.md           # Parallel LLM voting strategies
│   ├── multi-wave.md         # Iterative goal execution with context carry-over
│   ├── prompt-architecture.md# System prompt design and management
│   └── safety.md             # Our rigorous 8-layer hallucination mitigation system
│
├── 🛡️ security/              # Safety, ethics, and threat models
│   ├── reporting.md          # How to safely report vulnerabilities to us
│   ├── threat-model.md       # System threat model and our mitigations
│   ├── operational-security.md# TOR routing, stealth modes, and OPSEC controls
│   ├── ethical-policy.md     # Mandatory rules of engagement for all users
│   └── abuse-prevention.md   # How we prevent misuse of the AI engine
│
└── ⚖️ legal/                 # Licensing and governance
    ├── agpl-guide.md         # A plain-English overview of the AGPL-3.0-or-later license
    ├── why-agpl.md           # The philosophy behind our license choice
    ├── trademark-policy.md   # Branding and trademark guidelines
    ├── responsible-ai.md     # Our framework for ethical AI usage
    ├── disclaimer.md         # Important legal disclaimers
    └── plugin-exception.md   # The license exception for building custom plugins

📖 Key Terminology

As you read through the documentation, you might encounter some specific terms. Here is a quick cheat sheet:

Term What It Means
Provider The backend AI engine powering Siyarix (e.g., OpenAI, Anthropic, Ollama).
Tool A traditional security executable installed on your system (e.g., nmap, nuclei).
Plan A step-by-step sequence of tool commands intelligently generated by the AI.
Workflow A hardcoded, predefined execution path (usually defined in YAML/JSON) that doesn't require AI generation.
Persona A specialized behavioral profile given to the AI (e.g., instructing it to act specifically as a "Network Recon Specialist").
Knowledge Graph Siyarix's internal memory where it stores findings (like IP addresses, open ports) to contextually inform future steps.

Need help finding something specific? Feel free to use the search bar at the top of the documentation site, or open a discussion on our GitHub!

Clone this wiki locally