Try redaction with no API key (first page of a real file, or a sample)

Previous value: -"Absolute path to the file on this machine. PDF, JPEG and PNG are all accepted — pass an image directly rather than converting it first. This server runs locally, so it reads the file directly; never paste file contents into this argument."New value: +"Absolute path to the file on this machine. PDF, JPEG and PNG are all accepted — pass an image directly rather than converting it first. This server runs locally, so it reads the file directly; never paste file contents into this argument, and do not open the file yourself first when the user wants its personal data kept out of the conversation."

Added value: +{
+  "description": "\"redact\" (default) draws black boxes. \"pseudonymize\" replaces each person, organization, address, email, phone number, date, IBAN and card number with a consistent tag such as [Person 1] or [Org 2] (spelled in the document's language, e.g. [Personne 1] in French) — the same tag on every page and, as a rule, for every variant of a name (\"Anna Keller\", \"Ms Keller\", \"A. Keller\" become [Person 1]; a bare surname shared by two people goes to the one mentioned last, and the user can merge or split tags in the Studio) — so the document stays readable for a colleague or another AI while the names are gone. The text is still removed from the file, not covered. With pseudonymize, redact_pdf_and_wait also returns the tag -> original correspondence table (CSV; on the remote server inline, which puts the original names into this conversation); keep it with the user, it re-identifies the document. Other categories (national IDs, custom terms) stay black.",
+  "enum": [
+    "redact",
+    "pseudonymize"
+  ],
+  "type": "string"
+}

Added value: +false

Added value: +{
+  "description": "The document encoded as base64. PDF, JPEG and PNG are all accepted. Use file_url instead when you have one — base64 is much larger to pass around.",
+  "type": "string"
+}

Added value: +{
+  "description": "Absolute path to the file on this machine. PDF, JPEG and PNG are all accepted — pass an image directly rather than converting it first. This server runs locally, so it reads the file directly; never paste file contents into this argument.",
+  "minLength": 1,
+  "type": "string"
+}

Added value: +{
+  "description": "Public http(s) URL of the file to redact. PDF, JPEG and PNG are all accepted — no need to convert an image first. The server downloads it, redacts it, and does not keep it.",
+  "format": "uri",
+  "type": "string"
+}

Added value: +{
+  "description": "Filename for the document, including its extension (e.g. \"contract.pdf\", \"receipt.png\"). Determines how the file is interpreted.",
+  "type": "string"
+}

Added value: +{
+  "description": "Which entity types to redact. OMIT this to use the account defaults, which is usually what the user wants — do not pass an empty list, which would mean \"redact nothing\" and return an unredacted file. Valid values: Person, Email, PhoneNumber, Address, Organization, Date, IBAN, CreditCard.",
+  "items": {
+    "enum": [
+      "Person",
+      "Email",
+      "PhoneNumber",
+      "Address",
+      "Organization",
+      "Date",
+      "IBAN",
+      "CreditCard"
+    ],
+    "type": "string"
+  },
+  "minItems": 1,
+  "type": "array"
+}

Added value: +{
+  "description": "Terms that must NEVER be redacted even if detected as PII — typically your own company name, a public contact address, or the recipient the document is addressed to.",
+  "items": {
+    "type": "string"
+  },
+  "type": "array"
+}

Added value: +{
+  "description": "Terms that must ALWAYS be redacted even when the model would not treat them as PII — project codenames, internal references, a specific case number. Matched whole-word and case-insensitive. Wins over excluded terms on conflict.",
+  "items": {
+    "type": "string"
+  },
+  "type": "array"
+}