Architecture — Agentic Multi-Stage Bot
Description
A comprehensive guide to the system design, data flows, deployment topology, and extension points of the Agentic AI monorepo. ---
Installation
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open the source below and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
Repository README
This is the README for hoangsonww/Agentic-AI-Pipeline, shared by 3 entries
in this directory. It describes the repository, not this entry specifically.
Architecture — Agentic Multi-Stage Bot
A comprehensive guide to the system design, data flows, deployment topology, and extension points of the Agentic AI monorepo.
Table of Contents
- Executive Summary
- 7-Layer Architecture
- System Topology
- Core Agent Graph (LangGraph)
- Request Lifecycle
- Layer 1 — Experience
- Layer 2 — Discovery
- Layer 3 — Agent Composition
- Layer 4 — Reasoning & Planning
- Layer 5 — Tools & API
- Layer 6 — Memory & Feedback
- Layer 7 — Infrastructure
- MCP Server (Control Plane)
- Sub-Pipelines
- Social Media Automation
- Client SDKs
- Data Model
- Configuration
- Deployment Architecture
- Security Model
- Extension Points
- Directory Map
Executive Summary
The **Agentic Multi-Stage Bot** is a production-grade, multi-stage agentic system that orchestrates LLM reasoning, autonomous tool use, and persistent memory through a **7-layer architecture**. The reference implementation is a **DossierOutreachAgent** — given a topic or company, it builds a cited briefing and optionally drafts an outreach email.
The monorepo also ships three companion pipelines (RAG, Coding, Data), a shared MCP control plane, client SDKs (Python, TypeScript, .NET 8), and enterprise deployment modules for AWS, GCP, Azure, OCI, Kubernetes, Nomad, and bare-metal.
7-Layer Architecture
Each layer has a single responsibility and communicates through well-defined interfaces.
graph TB
subgraph L1["Layer 1 — Experience"]
FASTAPI["FastAPI + SSE"]
WEBUI["Vue.js Web UI"]
CLI["CLI (ingest / demo)"]
end
subgraph L2["Layer 2 — Discovery"]
WEBSEARCH["WebSearch\n(DuckDuckGo)"]
WEBFETCH["WebFetch\n(httpx + trafilatura)"]
KBSEARCH["KbSearch\n(ChromaDB)"]
end
subgraph L3["Layer 3 — Composition"]
PROFILE["AgentProfile\nDossierOutreachAgent"]
SYSTEM["System Prompt\n(SYSTEM constant)"]
end
subgraph L4["Layer 4 — Reasoning"]
PLAN["Plan"]
DECIDE["Decide"]
ACT["Act"]
TOOLS_NODE["ToolNode"]
REFLECT["Reflect"]
FINALIZE["Finalize"]
end
subgraph L5["Layer 5 — Tools"]
CALC["Calculator"]
FWRITE["FileWrite"]
EMAIL["Emailer"]
KB_ADD["KbAdd"]
SM["Social Media Tools"]
end
subgraph L6["Layer 6 — Memory"]
SQLITE[("SQLite\nmessages / feedback")]
CHROMA[("ChromaDB\nvector KB")]
end
subgraph L7["Layer 7 — Inf
Related Skills
mcp-server-postgres
Read-only PostgreSQL database access.
Data mcp-server-sqlite
SQLite database interaction and querying.
Data mcp-server-google-maps
Google Maps integration for location data.
Data Bitbucket Data Center
---
Data Csv Data Summarizer
Automatically analyze CSV files and generate comprehensive insights with visualizations
Data Financial Services
Reference agents, skills, and data connectors for the financial-services workflows we see most — investment ba
Data