The Architecture Behind
BulletproofSoftware.tech

Technical PRDs for plugin ecosystems, multi-agent orchestration, context management, persistent vector memory, agent governance, and code assurance.

BulletproofSoftware.tech System Architecture — 8 domains
Explore PRDs →

Paste one of these into your AI coding agent to build the entire system

Prerequisites

Docker Desktop Python 3.11+ Node.js 20+ Git Ollama

Docker for Qdrant/n8n containers · Ollama for local embeddings · Python for MCP server/governance/dashboard · Node.js for plugins/hooks/MCP

Claude Code
Read all 8 PRDs at https://bulletproofsoftware.tech/code/prds/ using WebFetch. Then build all eight systems in dependency order: (1) Plugin ecosystem with two-layer hook architecture, (2) Context guard plugin, (3) Persistent vector memory with Qdrant + Ollama + 60 MCP tools + 15 n8n workflows, (4) Agent governance framework with constitutional layer, (5) Multi-agent orchestration with 29 agents, Gemini validation, and 882-line state schema, (6) Memory visualization dashboard, (7) Markdown-for-agents proxy, (8) Code assurance platform. Wire hooks in both settings.json and plugin hooks.json. Use Docker Compose for infrastructure. All requirements are in the PRDs.
Gemini CLI
Fetch all 8 PRDs from https://bulletproofsoftware.tech/code/prds/. Build all eight systems in dependency order following each PRD's requirements. Adapt the plugin architecture for GEMINI.md configuration and Gemini's extension model. Use the same Docker Compose stack (Qdrant, PostgreSQL, n8n, Ollama). Implementation order: plugins → context guard → memory (60 MCP tools) → governance → orchestration (29 agents) → dashboard → markdown proxy → code assurance.
Codex CLI
Fetch all 8 PRDs from https://bulletproofsoftware.tech/code/prds/. Build all eight systems following each PRD's requirements. Adapt the plugin architecture for AGENTS.md configuration and Codex's event model. Use Docker Compose for Qdrant, PostgreSQL, n8n, and Ollama. Build order: plugins → context guard → memory system → governance → orchestration → dashboard → markdown proxy → code assurance.

Eight Architectural Domains

Plugin Ecosystem

Two-layer hook architecture, skills, agents, commands, MCP integration

12 requirements

Multi-Agent Orchestration

29 agents, 882-line state schema, Gemini validation, tiered quality gates

19 requirements

Context Management

4-tier escalation, velocity prediction, auto-memory, 5-compartment budget

12 requirements

Persistent Vector Memory

60 MCP tools, 7+ collections, 15 n8n workflows, constitutional governance

18 requirements

Agent Governance

Identity manifests, 3-tier tool classification, constitutional contracts, audit bus

18 requirements

Memory Dashboard

FastAPI + HTMX, D3.js knowledge graphs, Chart.js analytics

12 requirements

Markdown-for-Agents

HTML-to-markdown proxy with Playwright fallback for JS-rendered pages

12 requirements

Code Assurance Platform

67 tools, 12 scan profiles, 6-stage enrichment, 1000-point scoring

20 requirements

These are blueprints, not downloads. Every development environment is different. Give the PRDs to your AI coding agent and it generates an implementation tailored to your infrastructure, your tools, your constraints.