PramodGawali/ai-guardrail-platform

Clarify demo chat policy selection

Fix Hugging Face production database startup

Add enterprise guardrail admin and MCP controls

Add MCP and proxy integration surfaces

docs: update all docs for Python FastAPI implementation

feat: convert implementation from .NET to Python FastAPI

fix: add explicit app_port to HF Spaces metadata

fix: add Policy Admin nav link to demo page header

docs: add published benchmark results to README

fix: achieve 97.9% accuracy on 48-case eval suite

docs: rewrite README as a compelling project landing page

fix: detect DAN/jailbreak prompts in heuristic providers

feat: add policy admin UI for runtime rule management

docs: add in-house implementation plan and comprehensive 44-case test suite

fix: add destructive action and data exfiltration detection to heuristics

fix: word-boundary heuristic matching and risk score display

fix: use together provider with Qwen2.5-7B-Instruct-Turbo (verified working)

fix: use model-in-path router URL and switch to Qwen2.5-72B-Instruct

fix: migrate to router.huggingface.co, switch to Qwen2.5-7B-Instruct

fix: switch to Meta-Llama-3.2-1B-Instruct on standard inference URL

fix: switch to HF router URL and zephyr-7b-beta model

fix: switch to Mistral-7B and use correct HF serverless inference URL

fix: accept HF_TOKEN env var as fallback for HuggingFace token

docs: add guardrail examples for each check type

feat: add full AI chat loop with HuggingFace Inference API

resolve: merge conflict with HF default README

feat: Enterprise AI Guardrail Platform — HF Spaces ready

initial commit