Secure AI Customer Support Platform — AI skill for Claude Code
Secure LLM-powered customer support platform: an AI agent (Claude) with grounded RAG answers, authorization-enforced tool calling, human handoff and defence-in-depth against prompt injection.
How to install Secure AI Customer Support Platform
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open mojtaba-py-code/secure-ai-customer-support-platform and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Secure AI Customer Support Platform does
Secure LLM-powered customer support platform: an AI agent (Claude) with grounded RAG answers, authorization-enforced tool calling, human handoff and defence-in-depth against prompt injection. FastAPI, PostgreSQL, Qdrant, Redis - 480+ tests, fuzzed and scanned in CI.
Alternatives in AI
- AI Engineering Hub — In-depth tutorials on LLMs, RAGs and real-world AI agent applications 37.2k ★
- Empirica — Make AI agents and AI workflows measurably reliable 244 ★
- Karpathy LLM Wiki Stack — A comprehensive, build-ready reference for constructing a high-performance personal knowledge system using Obs 91 ★
README
AegisSupport AI
[](https://github.com/mojtaba-py-code/secure-ai-customer-support-platform/actions/workflows/ci.yml) [](https://github.com/mojtaba-py-code/secure-ai-customer-support-platform/actions/workflows/codeql.yml) [](https://scorecard.dev/viewer/?uri=github.com/mojtaba-py-code/secure-ai-customer-support-platform)  [](LICENSE)
**A secure, LLM-powered customer-support platform.** Customers chat with an AI assistant that answers from the company's knowledge base, looks up *their own* orders, payments, refunds and tickets through tightly controlled tools, prepares refunds and cancellations for the customer to confirm, and hands the conversation to a human agent whenever the situation calls for one.
The project treats the language model as an **untrusted component**. Every decision that matters for security - who may see which record, which tools a turn may use, whether a conversation needs a human, whether a reply is safe to show - is made in deterministic, tested code around the model, not by the model.
The demo business ("Acme Home Electronics"), its customers and orders are fictional. The platform runs fully offline with a deterministic stand-in model and local embeddings; switch one setting to use Claude.
Contents
- What it does
- Architecture
- Quick start (no Docker, no API key)
- Try it
- Using Claude
- [Docker Compose (production-like)](#docke
Related Skills
Refund Agent
AI customer-support agent that handles refund emails using Claude Opus 5.5 and tool calling. Refund rules are
RAG Support Agent
Financial document AI: evidence-checked extraction (structured outputs), RAG Q&A with page citations, a tool-c
AI Document Assistant
Secure, multi-tenant enterprise document intelligence & RAG platform: sandboxed ingestion, hybrid pgvector/Qdr
Frontend Design Pro
Machine-enforced frontend UI/UX skill pack for AI agents. Registry + lazy loading: a 2,149-token router loads
Customer Service
A multi-provider AI customer service agent that automates order lookup, return-eligibility checks, and Stripe
Vox Agent
LLM agent for customer support with inline evaluation, hallucination detection, retry/fallback logic, and RAG
Related Agents
LLM Sec Review
LLM/agent security review specialist — prompt injection, the Agents Rule of Two, tool-call authorization, mode
Appsec AI Security Reviewer
Use for AI/LLM application security review: direct/indirect prompt injection, tool and agent authorization, RA
Agentapp Builder
Agent 应用构建工程师(Context Engineering 合一)。当需要编写 system prompt / 角色设定、定义工具(tool/function calling)契约、设计 RAG 检索与 cont