{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreiej3h7cn4o5ruvj3lagvhw7fbjxb7bg37ffqtma72p77ogakfpy4m",
"uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3mmidjmwbjka2"
},
"path": "/t/promptshield-open-source-prompt-injection-hallucinationfirewall-for-any-llm-app/176174#post_1",
"publishedAt": "2026-05-23T01:27:13.000Z",
"site": "https://discuss.huggingface.co",
"tags": [
"Llmguard Demo - a Hugging Face Space by Neeraj-Ch0udhary",
"GitHub - Neeraj-Ch0udhary/promptshield: Open-source prompt injection firewall + hallucination blocker + agent memory layer for any LLM app · GitHub",
"Neeraj-Ch0udhary/llmguard-input-guard · Hugging Face"
],
"textContent": "Built a Python middleware that wraps any LLM call with 3 layers:\n\nInput Guard — blocks prompt injections (DistilBERT, F1: 0.944)\nOutput Guard — catches hallucinations (NLI cross-encoder)\nMemory Layer — persistent agent memory (ChromaDB)\n\nAll local models. No API keys. No cost.\n\npip install neeraj-llmguard\n\nLive demo: Llmguard Demo - a Hugging Face Space by Neeraj-Ch0udhary\nGitHub: GitHub - Neeraj-Ch0udhary/promptshield: Open-source prompt injection firewall + hallucination blocker + agent memory layer for any LLM app · GitHub\nModel: Neeraj-Ch0udhary/llmguard-input-guard · Hugging Face\n\nWould love feedback from the HF community!",
"title": "PromptShield — open-source prompt injection + hallucinationfirewall for any LLM app"
}