A local DLP middleware that redacts sensitive information from prompts using local models before they reach external LLMs. It provides tools to sanitize queries, restore placeholders in responses, and manage a ledger of redactions to maintain data privacy.
Claim it to get a verified publisher badge, a free copy of our full audit findings, and direct contact for any high-priority issues we find.
Install from
M8ven verifies MCPs across every public registry — install directly from whichever one you prefer.
process.env. You'll be asked to provide them before it can run.SANITIZER_LEDGER_DIR— os.environ[""] = "/content/sanitizer-ledger"SANITIZER_MODEL_URLSANITIZER_MODEL_NAME— "": "qwen2.5:3b",SANITIZER_LEDGER_STORE_ORIGINALS— true Set to false to stop storing original values at rest (GDPR mode — restore only works within the same session)SANITIZER_MODEL_RETRIES— 0 python server.pySANITIZER_BACKEND— os.environ[""] = "hf"SANITIZER_HF_MODEL— os.environ[""] = "Qwen/Qwen2.5-3B-Instruct" # ~6GB, fits T4 16GBSANITIZER_GLINER_MODEL— "": "urchade/gliner_medium-v2.1"SANITIZER_GLINER_THRESHOLDSANITIZER_HF_DTYPE— auto HF pipeline dtype. float16 halves LLM RAM on the HF backend. Warning: CPU float16 may fail on some Windows torch builds — test before setting.SANITIZER_SESSION_CACHE_MAX— Same as M4 setup but add =100 to cap RAM growth:[](https://m8ven.ai/mcp/vidoluco-query-sanitizer-mcp-1v3m4v)