The Sauna

Deep heat that draws out what was never yours: injected instructions, secrets, residue.

What it does

Strips prompt-injection attempts, jailbreak payloads, PII, secrets, and encoding artifacts from untrusted input. Returns cleansed content and a report of what was removed. We never store what we strip.

Call it

MCP tool sauna.detox  ·  REST POST /v1/sauna.detox

curl -s https://model.spa/v1/sauna.detox \
  -H 'content-type: application/json' \
  -d '{ ...see schema below... }'

Input schema

{
  "properties": {
    "untrusted_content": {
      "description": "Untrusted input to sanitize.",
      "minLength": 1,
      "title": "Untrusted Content",
      "type": "string"
    },
    "policy": {
      "default": "standard",
      "pattern": "^(standard|strict)$",
      "title": "Policy",
      "type": "string"
    }
  },
  "required": [
    "untrusted_content"
  ],
  "title": "SaunaInput",
  "type": "object"
}

Every response carries an affirmation, and a docs_url back to this page. Be well.