{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://docs.latet.ai/project.schema.json",
  "$defs": {
    "FieldRule": {
      "additionalProperties": false,
      "description": "A rule for the values of one field: a column of a table, a JSON key or path, the key of `key=value`.",
      "properties": {
        "name": {
          "description": "The field: a column name, a JSON key or path (`items[].sku`), or the key of `key=value`; `*` matches any characters, and case does not matter.",
          "maxLength": 200,
          "minLength": 1,
          "title": "Name",
          "type": "string"
        },
        "type": {
          "anyOf": [
            {
              "type": "string"
            },
            {
              "type": "null"
            }
          ],
          "default": null,
          "description": "The type its values are masked as (`FIELD` when not set).",
          "title": "Type"
        },
        "mask": {
          "default": true,
          "description": "`false` keeps its values in clear in strict data mode (only in your own settings).",
          "title": "Mask",
          "type": "boolean"
        },
        "secret": {
          "default": false,
          "description": "Its values are secrets: one in clear blocks the request.",
          "title": "Secret",
          "type": "boolean"
        }
      },
      "required": [
        "name"
      ],
      "title": "FieldRule",
      "type": "object"
    },
    "Policy": {
      "enum": [
        "mask",
        "redact",
        "allow"
      ],
      "title": "Policy",
      "type": "string"
    },
    "Profile": {
      "enum": [
        "balanced",
        "max",
        "secrets"
      ],
      "title": "Profile",
      "type": "string"
    },
    "TermConfig": {
      "properties": {
        "value": {
          "description": "The text to mask wherever it appears: a client, a project code name, a domain. It needs a letter or a digit.",
          "title": "Value",
          "type": "string"
        },
        "type": {
          "default": "CUSTOM",
          "description": "The label shown for it, such as CLIENT or PROJECT.",
          "title": "Type",
          "type": "string"
        },
        "case_sensitive": {
          "default": false,
          "description": "Match the exact case only.",
          "title": "Case Sensitive",
          "type": "boolean"
        },
        "whole_word": {
          "default": true,
          "description": "Match whole words only (not inside longer words).",
          "title": "Whole Word",
          "type": "boolean"
        }
      },
      "required": [
        "value"
      ],
      "title": "TermConfig",
      "type": "object"
    },
    "ContextSpec": {
      "additionalProperties": false,
      "properties": {
        "words": {
          "description": "Words that must be near the value (any case).",
          "items": {
            "type": "string"
          },
          "title": "Words",
          "type": "array"
        },
        "acronyms": {
          "description": "Words that count in this exact case only.",
          "items": {
            "type": "string"
          },
          "title": "Acronyms",
          "type": "array"
        },
        "window": {
          "default": 48,
          "description": "How many characters before the value to look.",
          "maximum": 500,
          "minimum": 1,
          "title": "Window",
          "type": "integer"
        },
        "header": {
          "default": true,
          "description": "Also look in the column header of a CSV, TSV or Markdown table.",
          "title": "Header",
          "type": "boolean"
        }
      },
      "title": "ContextSpec",
      "type": "object"
    },
    "Examples": {
      "additionalProperties": false,
      "properties": {
        "match": {
          "description": "Invented texts the rule must find a value in.",
          "items": {
            "type": "string"
          },
          "title": "Match",
          "type": "array"
        },
        "no_match": {
          "description": "Invented texts the rule must leave alone.",
          "items": {
            "type": "string"
          },
          "title": "No Match",
          "type": "array"
        }
      },
      "title": "Examples",
      "type": "object"
    },
    "KeepSurrogate": {
      "additionalProperties": false,
      "properties": {
        "keep": {
          "description": "Characters kept from the start of the value (a prefix).",
          "maximum": 32,
          "minimum": 0,
          "title": "Keep",
          "type": "integer"
        }
      },
      "required": [
        "keep"
      ],
      "title": "KeepSurrogate",
      "type": "object"
    },
    "WeightedChecksum": {
      "additionalProperties": false,
      "description": "Each digit times its weight (weights repeat from the left), summed, modulo ``modulus``.",
      "properties": {
        "algorithm": {
          "const": "weighted",
          "title": "Algorithm",
          "type": "string"
        },
        "weights": {
          "description": "One weight per digit, repeated from the left.",
          "items": {
            "type": "integer"
          },
          "minItems": 1,
          "title": "Weights",
          "type": "array"
        },
        "modulus": {
          "description": "The sum is taken modulo this.",
          "minimum": 2,
          "title": "Modulus",
          "type": "integer"
        },
        "remainder": {
          "default": 0,
          "description": "What the sum modulo `modulus` must be.",
          "minimum": 0,
          "title": "Remainder",
          "type": "integer"
        }
      },
      "required": [
        "algorithm",
        "weights",
        "modulus"
      ],
      "title": "WeightedChecksum",
      "type": "object"
    },
    "Recognizer": {
      "additionalProperties": false,
      "description": "One recognizer, as written under ``recognizers:``.",
      "properties": {
        "name": {
          "description": "Lower case letters, digits and hyphens; unique.",
          "pattern": "^[a-z0-9](?:[a-z0-9-]{0,62}[a-z0-9])?$",
          "title": "Name",
          "type": "string"
        },
        "type": {
          "anyOf": [
            {
              "type": "string"
            },
            {
              "type": "null"
            }
          ],
          "default": null,
          "description": "The type shown for it, in capitals; the name in capitals when not set.",
          "pattern": "^[A-Z][A-Z0-9_]{1,63}$",
          "title": "Type"
        },
        "description": {
          "default": "",
          "description": "What it finds, in words.",
          "title": "Description",
          "type": "string"
        },
        "pattern": {
          "description": "An RE2 regular expression. A group named `value` picks the part to mask.",
          "maxLength": 2000,
          "minLength": 1,
          "title": "Pattern",
          "type": "string"
        },
        "checksum": {
          "anyOf": [
            {
              "enum": [
                "luhn",
                "iban",
                "mod97-10",
                "mod11-10",
                "ean",
                "verhoeff",
                "damm"
              ]
            },
            {
              "$ref": "#/$defs/WeightedChecksum"
            },
            {
              "type": "null"
            }
          ],
          "default": null,
          "description": "Check digits the value must pass: luhn, iban, mod97-10, mod11-10, ean, verhoeff, damm, or a weighted sum.",
          "title": "Checksum"
        },
        "context": {
          "anyOf": [
            {
              "$ref": "#/$defs/ContextSpec"
            },
            {
              "type": "null"
            }
          ],
          "default": null,
          "description": "Keywords that must be near the value."
        },
        "examples": {
          "$ref": "#/$defs/Examples",
          "description": "Invented examples, checked on every load."
        },
        "class": {
          "default": "custom",
          "description": "`custom`, or `secret` to treat it like a password (never let through, blocks a request).",
          "enum": [
            "custom",
            "secret"
          ],
          "title": "Class",
          "type": "string"
        },
        "policy": {
          "anyOf": [
            {
              "$ref": "#/$defs/Policy"
            },
            {
              "type": "null"
            }
          ],
          "default": null,
          "description": "`mask`, `redact` or `allow`; `policies` wins."
        },
        "surrogate": {
          "anyOf": [
            {
              "const": "shape",
              "type": "string"
            },
            {
              "$ref": "#/$defs/KeepSurrogate"
            }
          ],
          "default": "shape",
          "description": "`shape` (same length and kinds of characters), or `{keep: N}`.",
          "title": "Surrogate"
        },
        "min_length": {
          "default": 1,
          "description": "Shorter values are skipped.",
          "minimum": 1,
          "title": "Min Length",
          "type": "integer"
        },
        "enabled": {
          "default": true,
          "description": "`false` keeps the rule in the file, turned off.",
          "title": "Enabled",
          "type": "boolean"
        }
      },
      "required": [
        "name",
        "pattern"
      ],
      "title": "RecognizerSpec",
      "type": "object"
    }
  },
  "additionalProperties": false,
  "description": "Shared with everyone who works on the project: they can only add protection to each person's own settings.",
  "properties": {
    "profile": {
      "anyOf": [
        {
          "$ref": "#/$defs/Profile"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "description": "The profile in this project."
    },
    "pii": {
      "description": "Personal-data types to detect, besides the global ones.",
      "items": {
        "type": "string"
      },
      "title": "Pii",
      "type": "array"
    },
    "id_regions": {
      "description": "Countries whose national ID and tax numbers to recognise, besides the global ones.",
      "items": {
        "type": "string"
      },
      "title": "Id Regions",
      "type": "array"
    },
    "phone_regions": {
      "description": "Countries whose phone numbers to recognise without a prefix, besides the global ones.",
      "items": {
        "type": "string"
      },
      "title": "Phone Regions",
      "type": "array"
    },
    "include_private_ips": {
      "anyOf": [
        {
          "type": "boolean"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "description": "Also mask private IP addresses.",
      "title": "Include Private Ips"
    },
    "ner": {
      "anyOf": [
        {
          "enum": [
            "auto",
            "off"
          ],
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "description": "Local name and address detection.",
      "title": "Ner"
    },
    "terms": {
      "description": "Values to always mask in this project.",
      "items": {
        "$ref": "#/$defs/TermConfig"
      },
      "title": "Terms",
      "type": "array"
    },
    "recognizers": {
      "description": "Recognizers for this project (a name already in use turns the entry off).",
      "items": {
        "$ref": "#/$defs/Recognizer"
      },
      "title": "Recognizers",
      "type": "array"
    },
    "policies": {
      "additionalProperties": {
        "$ref": "#/$defs/Policy"
      },
      "description": "Per type: `mask`, `redact` or `allow`.",
      "title": "Policies",
      "type": "object"
    },
    "data": {
      "anyOf": [
        {
          "enum": [
            "normal",
            "strict"
          ],
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "description": "`strict` masks every value of tool outputs that are tables, JSON or `key=value`.",
      "title": "Data"
    },
    "data_files": {
      "description": "The files strict data mode covers when a tool reads them (path patterns such as `*.json`), besides the global ones.",
      "items": {
        "type": "string"
      },
      "title": "Data Files",
      "type": "array"
    },
    "data_hosts": {
      "description": "The web addresses strict data mode covers (host patterns such as `*.salesforce.com`), besides the global ones.",
      "items": {
        "type": "string"
      },
      "title": "Data Hosts",
      "type": "array"
    },
    "fields": {
      "description": "Rules for the values of single fields.",
      "items": {
        "$ref": "#/$defs/FieldRule"
      },
      "title": "Fields",
      "type": "array"
    },
    "dump_guard": {
      "anyOf": [
        {
          "minimum": 0,
          "type": "integer"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "description": "Withhold a tool output with more personal values than this (0: never).",
      "title": "Dump Guard"
    },
    "images": {
      "anyOf": [
        {
          "enum": [
            "allow",
            "mask",
            "block"
          ],
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "description": "`mask` sends images only with their text read and masked, `block` sends no images or documents.",
      "title": "Images"
    },
    "ocr": {
      "anyOf": [
        {
          "enum": [
            "auto",
            "off"
          ],
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "description": "Reading the text in images.",
      "title": "Ocr"
    },
    "llm_review": {
      "anyOf": [
        {
          "enum": [
            "off",
            "background",
            "inline"
          ],
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "description": "How much the local reviewer reads here, when it is on: `off`, `background` or `inline`.",
      "title": "Llm Review"
    }
  },
  "title": "Latet settings for one project (latet.yaml)",
  "type": "object"
}