dsh-data-quality 0.1.3 → 0.3.0

package/lib/types/tools/profile-report-schema.d.ts

@@ -0,0 +1,319 @@
+/**
+ * Shared JSON-Schema spec for the profile report's canonical value, reused by
+ * `data_profile` (its output) and `data_clean` (the `diffPreview` before/after
+ * snapshots). One source of truth so the two tools never drift.
+ * @module dsh-data-quality/tools-profile-report-schema
+ */
+/** One column card's schema. */
+export declare const COLUMN_PROFILE_SCHEMA: {
+    readonly type: "object";
+    readonly properties: {
+        readonly name: {
+            readonly type: "string";
+            readonly required: true;
+        };
+        readonly inferredType: {
+            readonly type: "string";
+            readonly enum: readonly ["number", "date", "boolean", "string", "empty", "mixed"];
+            readonly required: true;
+        };
+        readonly missing: {
+            readonly type: "number";
+            readonly required: true;
+        };
+        readonly missingRate: {
+            readonly type: "number";
+            readonly required: true;
+        };
+        readonly unique: {
+            readonly type: "number";
+            readonly required: true;
+        };
+        readonly numeric: {
+            readonly type: "object";
+            readonly properties: {
+                readonly count: {
+                    readonly type: "number";
+                    readonly required: true;
+                };
+                readonly distinct: {
+                    readonly type: "number";
+                    readonly required: true;
+                };
+                readonly min: {
+                    readonly type: "number";
+                    readonly required: true;
+                };
+                readonly max: {
+                    readonly type: "number";
+                    readonly required: true;
+                };
+                readonly mean: {
+                    readonly type: "number";
+                    readonly required: true;
+                };
+                readonly median: {
+                    readonly type: "number";
+                    readonly required: true;
+                };
+                readonly p25: {
+                    readonly type: "number";
+                    readonly required: true;
+                };
+                readonly p75: {
+                    readonly type: "number";
+                    readonly required: true;
+                };
+                readonly outliers: {
+                    readonly type: "number";
+                    readonly required: true;
+                };
+            };
+            readonly additionalProperties: false;
+        };
+        readonly topValues: {
+            readonly type: "array";
+            readonly items: {
+                readonly type: "object";
+                readonly properties: {
+                    readonly value: {
+                        readonly type: "string";
+                        readonly required: true;
+                    };
+                    readonly count: {
+                        readonly type: "number";
+                        readonly required: true;
+                    };
+                };
+                readonly additionalProperties: false;
+            };
+        };
+        readonly notes: {
+            readonly type: "array";
+            readonly items: {
+                readonly type: "string";
+            };
+            readonly required: true;
+        };
+    };
+    readonly additionalProperties: false;
+};
+/** The full profile report's schema. */
+export declare const PROFILE_REPORT_SCHEMA: {
+    readonly type: "object";
+    readonly properties: {
+        readonly schemaVersion: {
+            readonly type: "number";
+            readonly required: true;
+        };
+        readonly dataset: {
+            readonly type: "string";
+            readonly required: true;
+        };
+        readonly rowCount: {
+            readonly type: "number";
+            readonly required: true;
+        };
+        readonly sampled: {
+            readonly type: "boolean";
+            readonly required: true;
+        };
+        readonly profiledRows: {
+            readonly type: "number";
+            readonly required: true;
+        };
+        readonly columnCount: {
+            readonly type: "number";
+            readonly required: true;
+        };
+        readonly duplicateRows: {
+            readonly type: "number";
+            readonly required: true;
+        };
+        readonly duplicateRate: {
+            readonly type: "number";
+            readonly required: true;
+        };
+        readonly duplicateSampleRowIndexes: {
+            readonly type: "array";
+            readonly items: {
+                readonly type: "number";
+            };
+            readonly required: true;
+        };
+        readonly scorecard: {
+            readonly type: "object";
+            readonly properties: {
+                readonly overall: {
+                    readonly oneOf: readonly [{
+                        readonly type: "number";
+                    }, {
+                        readonly type: "null";
+                    }];
+                    readonly required: true;
+                };
+                readonly weightedOverall: {
+                    readonly oneOf: readonly [{
+                        readonly type: "number";
+                    }, {
+                        readonly type: "null";
+                    }];
+                    readonly required: true;
+                };
+                readonly dimensions: {
+                    readonly type: "array";
+                    readonly items: {
+                        readonly type: "object";
+                        readonly properties: {
+                            readonly name: {
+                                readonly type: "string";
+                                readonly enum: readonly ["completeness", "uniqueness", "validity", "consistency", "timeliness", "accuracy"];
+                                readonly required: true;
+                            };
+                            readonly score: {
+                                readonly oneOf: readonly [{
+                                    readonly type: "number";
+                                }, {
+                                    readonly type: "null";
+                                }];
+                                readonly required: true;
+                            };
+                            readonly note: {
+                                readonly type: "string";
+                                readonly required: true;
+                            };
+                        };
+                        readonly additionalProperties: false;
+                    };
+                    readonly required: true;
+                };
+            };
+            readonly additionalProperties: false;
+            readonly required: true;
+        };
+        readonly encoding: {
+            readonly type: "object";
+            readonly properties: {
+                readonly bom: {
+                    readonly oneOf: readonly [{
+                        readonly type: "string";
+                    }, {
+                        readonly type: "null";
+                    }];
+                    readonly required: true;
+                };
+                readonly validUtf8: {
+                    readonly type: "boolean";
+                    readonly required: true;
+                };
+            };
+            readonly additionalProperties: false;
+        };
+        readonly generatedAt: {
+            readonly type: "number";
+            readonly required: true;
+        };
+        readonly reportKey: {
+            readonly type: "string";
+        };
+        readonly columns: {
+            readonly type: "array";
+            readonly items: {
+                readonly type: "object";
+                readonly properties: {
+                    readonly name: {
+                        readonly type: "string";
+                        readonly required: true;
+                    };
+                    readonly inferredType: {
+                        readonly type: "string";
+                        readonly enum: readonly ["number", "date", "boolean", "string", "empty", "mixed"];
+                        readonly required: true;
+                    };
+                    readonly missing: {
+                        readonly type: "number";
+                        readonly required: true;
+                    };
+                    readonly missingRate: {
+                        readonly type: "number";
+                        readonly required: true;
+                    };
+                    readonly unique: {
+                        readonly type: "number";
+                        readonly required: true;
+                    };
+                    readonly numeric: {
+                        readonly type: "object";
+                        readonly properties: {
+                            readonly count: {
+                                readonly type: "number";
+                                readonly required: true;
+                            };
+                            readonly distinct: {
+                                readonly type: "number";
+                                readonly required: true;
+                            };
+                            readonly min: {
+                                readonly type: "number";
+                                readonly required: true;
+                            };
+                            readonly max: {
+                                readonly type: "number";
+                                readonly required: true;
+                            };
+                            readonly mean: {
+                                readonly type: "number";
+                                readonly required: true;
+                            };
+                            readonly median: {
+                                readonly type: "number";
+                                readonly required: true;
+                            };
+                            readonly p25: {
+                                readonly type: "number";
+                                readonly required: true;
+                            };
+                            readonly p75: {
+                                readonly type: "number";
+                                readonly required: true;
+                            };
+                            readonly outliers: {
+                                readonly type: "number";
+                                readonly required: true;
+                            };
+                        };
+                        readonly additionalProperties: false;
+                    };
+                    readonly topValues: {
+                        readonly type: "array";
+                        readonly items: {
+                            readonly type: "object";
+                            readonly properties: {
+                                readonly value: {
+                                    readonly type: "string";
+                                    readonly required: true;
+                                };
+                                readonly count: {
+                                    readonly type: "number";
+                                    readonly required: true;
+                                };
+                            };
+                            readonly additionalProperties: false;
+                        };
+                    };
+                    readonly notes: {
+                        readonly type: "array";
+                        readonly items: {
+                            readonly type: "string";
+                        };
+                        readonly required: true;
+                    };
+                };
+                readonly additionalProperties: false;
+            };
+            readonly required: true;
+        };
+    };
+    readonly additionalProperties: false;
+};
+//# sourceMappingURL=profile-report-schema.d.ts.map

package/lib/types/tools/profile-report-schema.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"profile-report-schema.d.ts","sourceRoot":"","sources":["../../../src/tools/profile-report-schema.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAIH,gCAAgC;AAChC,eAAO,MAAM,qBAAqB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;CAqCxB,CAAA;AAEV,wCAAwC;AACxC,eAAO,MAAM,qBAAqB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;CA+CxB,CAAA"}

package/lib/types/tools/profile-report-schema.js

@@ -0,0 +1,96 @@
+/**
+ * Shared JSON-Schema spec for the profile report's canonical value, reused by
+ * `data_profile` (its output) and `data_clean` (the `diffPreview` before/after
+ * snapshots). One source of truth so the two tools never drift.
+ * @module dsh-data-quality/tools-profile-report-schema
+ */
+const SCORECARD_DIMENSION_NAMES = ['completeness', 'uniqueness', 'validity', 'consistency', 'timeliness', 'accuracy'];
+/** One column card's schema. */
+export const COLUMN_PROFILE_SCHEMA = {
+    type: 'object',
+    properties: {
+        name: { type: 'string', required: true },
+        inferredType: { type: 'string', enum: ['number', 'date', 'boolean', 'string', 'empty', 'mixed'], required: true },
+        missing: { type: 'number', required: true },
+        missingRate: { type: 'number', required: true },
+        unique: { type: 'number', required: true },
+        numeric: {
+            type: 'object',
+            properties: {
+                count: { type: 'number', required: true },
+                distinct: { type: 'number', required: true },
+                min: { type: 'number', required: true },
+                max: { type: 'number', required: true },
+                mean: { type: 'number', required: true },
+                median: { type: 'number', required: true },
+                p25: { type: 'number', required: true },
+                p75: { type: 'number', required: true },
+                outliers: { type: 'number', required: true },
+            },
+            additionalProperties: false,
+        },
+        topValues: {
+            type: 'array',
+            items: {
+                type: 'object',
+                properties: {
+                    value: { type: 'string', required: true },
+                    count: { type: 'number', required: true },
+                },
+                additionalProperties: false,
+            },
+        },
+        notes: { type: 'array', items: { type: 'string' }, required: true },
+    },
+    additionalProperties: false,
+};
+/** The full profile report's schema. */
+export const PROFILE_REPORT_SCHEMA = {
+    type: 'object',
+    properties: {
+        schemaVersion: { type: 'number', required: true },
+        dataset: { type: 'string', required: true },
+        rowCount: { type: 'number', required: true },
+        sampled: { type: 'boolean', required: true },
+        profiledRows: { type: 'number', required: true },
+        columnCount: { type: 'number', required: true },
+        duplicateRows: { type: 'number', required: true },
+        duplicateRate: { type: 'number', required: true },
+        duplicateSampleRowIndexes: { type: 'array', items: { type: 'number' }, required: true },
+        scorecard: {
+            type: 'object',
+            properties: {
+                overall: { oneOf: [{ type: 'number' }, { type: 'null' }], required: true },
+                weightedOverall: { oneOf: [{ type: 'number' }, { type: 'null' }], required: true },
+                dimensions: {
+                    type: 'array',
+                    items: {
+                        type: 'object',
+                        properties: {
+                            name: { type: 'string', enum: [...SCORECARD_DIMENSION_NAMES], required: true },
+                            score: { oneOf: [{ type: 'number' }, { type: 'null' }], required: true },
+                            note: { type: 'string', required: true },
+                        },
+                        additionalProperties: false,
+                    },
+                    required: true,
+                },
+            },
+            additionalProperties: false,
+            required: true,
+        },
+        encoding: {
+            type: 'object',
+            properties: {
+                bom: { oneOf: [{ type: 'string' }, { type: 'null' }], required: true },
+                validUtf8: { type: 'boolean', required: true },
+            },
+            additionalProperties: false,
+        },
+        generatedAt: { type: 'number', required: true },
+        reportKey: { type: 'string' },
+        columns: { type: 'array', items: COLUMN_PROFILE_SCHEMA, required: true },
+    },
+    additionalProperties: false,
+};
+//# sourceMappingURL=profile-report-schema.js.map

package/lib/types/tools/profile-report-schema.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"profile-report-schema.js","sourceRoot":"","sources":["../../../src/tools/profile-report-schema.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,MAAM,yBAAyB,GAAG,CAAC,cAAc,EAAE,YAAY,EAAE,UAAU,EAAE,aAAa,EAAE,YAAY,EAAE,UAAU,CAAU,CAAA;AAE9H,gCAAgC;AAChC,MAAM,CAAC,MAAM,qBAAqB,GAAG;IACnC,IAAI,EAAE,QAAQ;IACd,UAAU,EAAE;QACV,IAAI,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;QACxC,YAAY,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,QAAQ,EAAE,MAAM,EAAE,SAAS,EAAE,QAAQ,EAAE,OAAO,EAAE,OAAO,CAAC,EAAE,QAAQ,EAAE,IAAI,EAAE;QACjH,OAAO,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;QAC3C,WAAW,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;QAC/C,MAAM,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;QAC1C,OAAO,EAAE;YACP,IAAI,EAAE,QAAQ;YACd,UAAU,EAAE;gBACV,KAAK,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;gBACzC,QAAQ,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;gBAC5C,GAAG,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;gBACvC,GAAG,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;gBACvC,IAAI,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;gBACxC,MAAM,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;gBAC1C,GAAG,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;gBACvC,GAAG,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;gBACvC,QAAQ,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;aAC7C;YACD,oBAAoB,EAAE,KAAK;SAC5B;QACD,SAAS,EAAE;YACT,IAAI,EAAE,OAAO;YACb,KAAK,EAAE;gBACL,IAAI,EAAE,QAAQ;gBACd,UAAU,EAAE;oBACV,KAAK,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;oBACzC,KAAK,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;iBAC1C;gBACD,oBAAoB,EAAE,KAAK;aAC5B;SACF;QACD,KAAK,EAAE,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,EAAE,QAAQ,EAAE,IAAI,EAAE;KACpE;IACD,oBAAoB,EAAE,KAAK;CACnB,CAAA;AAEV,wCAAwC;AACxC,MAAM,CAAC,MAAM,qBAAqB,GAAG;IACnC,IAAI,EAAE,QAAQ;IACd,UAAU,EAAE;QACV,aAAa,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;QACjD,OAAO,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;QAC3C,QAAQ,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;QAC5C,OAAO,EAAE,EAAE,IAAI,EAAE,SAAS,EAAE,QAAQ,EAAE,IAAI,EAAE;QAC5C,YAAY,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;QAChD,WAAW,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;QAC/C,aAAa,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;QACjD,aAAa,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;QACjD,yBAAyB,EAAE,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,EAAE,QAAQ,EAAE,IAAI,EAAE;QACvF,SAAS,EAAE;YACT,IAAI,EAAE,QAAQ;YACd,UAAU,EAAE;gBACV,OAAO,EAAE,EAAE,KAAK,EAAE,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,EAAE,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC,EAAE,QAAQ,EAAE,IAAI,EAAE;gBAC1E,eAAe,EAAE,EAAE,KAAK,EAAE,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,EAAE,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC,EAAE,QAAQ,EAAE,IAAI,EAAE;gBAClF,UAAU,EAAE;oBACV,IAAI,EAAE,OAAO;oBACb,KAAK,EAAE;wBACL,IAAI,EAAE,QAAQ;wBACd,UAAU,EAAE;4BACV,IAAI,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,GAAG,yBAAyB,CAAC,EAAE,QAAQ,EAAE,IAAI,EAAE;4BAC9E,KAAK,EAAE,EAAE,KAAK,EAAE,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,EAAE,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC,EAAE,QAAQ,EAAE,IAAI,EAAE;4BACxE,IAAI,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;yBACzC;wBACD,oBAAoB,EAAE,KAAK;qBAC5B;oBACD,QAAQ,EAAE,IAAI;iBACf;aACF;YACD,oBAAoB,EAAE,KAAK;YAC3B,QAAQ,EAAE,IAAI;SACf;QACD,QAAQ,EAAE;YACR,IAAI,EAAE,QAAQ;YACd,UAAU,EAAE;gBACV,GAAG,EAAE,EAAE,KAAK,EAAE,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,EAAE,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC,EAAE,QAAQ,EAAE,IAAI,EAAE;gBACtE,SAAS,EAAE,EAAE,IAAI,EAAE,SAAS,EAAE,QAAQ,EAAE,IAAI,EAAE;aAC/C;YACD,oBAAoB,EAAE,KAAK;SAC5B;QACD,WAAW,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;QAC/C,SAAS,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE;QAC7B,OAAO,EAAE,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,qBAAqB,EAAE,QAAQ,EAAE,IAAI,EAAE;KACzE;IACD,oBAAoB,EAAE,KAAK;CACnB,CAAA"}

package/lib/types/tools/profile.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"profile.d.ts","sourceRoot":"","sources":["../../../src/tools/profile.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAGH,OAAO,KAAK,EAAE,kBAAkB,EAAE,MAAM,eAAe,CAAA;AAyCvD;;;;GAIG;AACH,wBAAgB,iBAAiB,CAAC,OAAO,EAAE,kBAAkB,mDAwC5D"}
+{"version":3,"file":"profile.d.ts","sourceRoot":"","sources":["../../../src/tools/profile.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAGH,OAAO,KAAK,EAAE,kBAAkB,EAAE,MAAM,eAAe,CAAA;AAKvD;;;;GAIG;AACH,wBAAgB,iBAAiB,CAAC,OAAO,EAAE,kBAAkB,mDA4B5D"}

package/lib/types/tools/profile.js

@@ -5,43 +5,8 @@
  */
 import { defineTool } from '@deepseek-ai/dsh-tools';
 import { renderProfileText } from "../profile.js";
+import { PROFILE_REPORT_SCHEMA } from "./profile-report-schema.js";
 import { workspaceOf } from "./shared.js";
-const COLUMN_PROFILE_SCHEMA = {
-    type: 'object',
-    properties: {
-        name: { type: 'string', required: true },
-        inferredType: { type: 'string', enum: ['number', 'date', 'boolean', 'string', 'empty', 'mixed'], required: true },
-        missing: { type: 'number', required: true },
-        missingRate: { type: 'number', required: true },
-        unique: { type: 'number', required: true },
-        numeric: {
-            type: 'object',
-            properties: {
-                min: { type: 'number', required: true },
-                max: { type: 'number', required: true },
-                mean: { type: 'number', required: true },
-                median: { type: 'number', required: true },
-                p25: { type: 'number', required: true },
-                p75: { type: 'number', required: true },
-                outliers: { type: 'number', required: true },
-            },
-            additionalProperties: false,
-        },
-        topValues: {
-            type: 'array',
-            items: {
-                type: 'object',
-                properties: {
-                    value: { type: 'string', required: true },
-                    count: { type: 'number', required: true },
-                },
-                additionalProperties: false,
-            },
-        },
-        notes: { type: 'array', items: { type: 'string' }, required: true },
-    },
-    additionalProperties: false,
-};
 /**
  * Build the `data_profile` tool definition against a mounted service.
  * @param service - the mounted ctx.dataQuality implementation.
@@ -52,35 +17,23 @@ export function defineProfileTool(service) {
         name: 'data_profile',
         description: [
             'Profile a workspace CSV/TSV/JSON/JSONL dataset with deterministic TypeScript computation (no mental math).',
-            'Returns row/column counts, inferred column types, missing rates, unique counts, numeric distributions (min/max/mean/median/p25/p75), IQR outlier counts, mixed-type suspicion notes, and duplicate-row counts.',
-            'Column cards cover every row by default; pass sample for a deterministic systematic sample on large files. Datasets above the configured row/size caps are rejected — use sample or raise the caps. The full report persists to the data_quality storage domain (reportKey in the result).',
+            'Returns row/column counts, inferred column types, missing rates, unique counts, numeric distributions (count/distinct/min/max/mean/median/p25/p75), IQR outlier counts, mixed-type suspicion notes, sha256 duplicate-row detection (rate + sample indexes), file encoding (BOM/UTF-8 validity), and a weighted DAMA six-dimension scorecard.',
+            'Pass industryPreset (retail/saas/fund/real-estate/e-commerce/healthcare/logistics/manufacturing/energy) to compare the dataset against that industry\'s expected columns, making the scorecard accuracy dimension determinable. Column cards cover every row by default; pass sample for a deterministic systematic sample on large files. Datasets above the configured row/size caps are rejected — use sample or raise the caps. The full report persists to the data_quality storage domain (reportKey in the result).',
         ].join('\n'),
         parameters: {
             path: { type: 'string', required: true, description: 'Workspace-relative dataset path (.csv/.tsv/.json/.jsonl). JSON datasets must be an array of flat objects.' },
             sample: { type: 'number', description: 'Optional systematic sample size (every ceil(N/sample)-th row) for the column cards; row counts stay exact.' },
+            industryPreset: { type: 'string', enum: ['retail', 'saas', 'fund', 'real-estate', 'e-commerce', 'healthcare', 'logistics', 'manufacturing', 'energy'], description: 'Optional industry preset id; its expected columns feed the scorecard accuracy dimension.' },
         },
         output: {
-            schema: {
-                type: 'object',
-                properties: {
-                    dataset: { type: 'string', required: true },
-                    rowCount: { type: 'number', required: true },
-                    sampled: { type: 'boolean', required: true },
-                    profiledRows: { type: 'number', required: true },
-                    columnCount: { type: 'number', required: true },
-                    duplicateRows: { type: 'number', required: true },
-                    generatedAt: { type: 'number', required: true },
-                    reportKey: { type: 'string' },
-                    columns: { type: 'array', items: COLUMN_PROFILE_SCHEMA, required: true },
-                },
-                additionalProperties: false,
-            },
+            schema: PROFILE_REPORT_SCHEMA,
             render: (_args, value) => [{ type: 'text', text: renderProfileText(value) }],
         },
         async execute(args, exec) {
             return service.profileDataset({
                 dataset: args.path,
                 sample: args.sample,
+                ...(args.industryPreset !== undefined ? { industryPreset: args.industryPreset } : {}),
                 workspace: workspaceOf(exec),
                 session: exec.agent?.session,
                 signal: exec.signal,

package/lib/types/tools/profile.js.map

@@ -1 +1 @@
-{"version":3,"file":"profile.js","sourceRoot":"","sources":["../../../src/tools/profile.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,UAAU,EAAE,MAAM,wBAAwB,CAAA;AAEnD,OAAO,EAAE,iBAAiB,EAAE,MAAM,eAAe,CAAA;AACjD,OAAO,EAAE,WAAW,EAAE,MAAM,aAAa,CAAA;AAEzC,MAAM,qBAAqB,GAAG;IAC5B,IAAI,EAAE,QAAQ;IACd,UAAU,EAAE;QACV,IAAI,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;QACxC,YAAY,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,QAAQ,EAAE,MAAM,EAAE,SAAS,EAAE,QAAQ,EAAE,OAAO,EAAE,OAAO,CAAC,EAAE,QAAQ,EAAE,IAAI,EAAE;QACjH,OAAO,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;QAC3C,WAAW,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;QAC/C,MAAM,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;QAC1C,OAAO,EAAE;YACP,IAAI,EAAE,QAAQ;YACd,UAAU,EAAE;gBACV,GAAG,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;gBACvC,GAAG,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;gBACvC,IAAI,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;gBACxC,MAAM,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;gBAC1C,GAAG,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;gBACvC,GAAG,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;gBACvC,QAAQ,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;aAC7C;YACD,oBAAoB,EAAE,KAAK;SAC5B;QACD,SAAS,EAAE;YACT,IAAI,EAAE,OAAO;YACb,KAAK,EAAE;gBACL,IAAI,EAAE,QAAQ;gBACd,UAAU,EAAE;oBACV,KAAK,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;oBACzC,KAAK,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;iBAC1C;gBACD,oBAAoB,EAAE,KAAK;aAC5B;SACF;QACD,KAAK,EAAE,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,EAAE,QAAQ,EAAE,IAAI,EAAE;KACpE;IACD,oBAAoB,EAAE,KAAK;CACnB,CAAA;AAEV;;;;GAIG;AACH,MAAM,UAAU,iBAAiB,CAAC,OAA2B;IAC3D,OAAO,UAAU,CAAC;QAChB,IAAI,EAAE,cAAc;QACpB,WAAW,EAAE;YACX,4GAA4G;YAC5G,gNAAgN;YAChN,4RAA4R;SAC7R,CAAC,IAAI,CAAC,IAAI,CAAC;QACZ,UAAU,EAAE;YACV,IAAI,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE,WAAW,EAAE,2GAA2G,EAAE;YAClK,MAAM,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,WAAW,EAAE,4GAA4G,EAAE;SACtJ;QACD,MAAM,EAAE;YACN,MAAM,EAAE;gBACN,IAAI,EAAE,QAAQ;gBACd,UAAU,EAAE;oBACV,OAAO,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;oBAC3C,QAAQ,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;oBAC5C,OAAO,EAAE,EAAE,IAAI,EAAE,SAAS,EAAE,QAAQ,EAAE,IAAI,EAAE;oBAC5C,YAAY,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;oBAChD,WAAW,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;oBAC/C,aAAa,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;oBACjD,WAAW,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE;oBAC/C,SAAS,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE;oBAC7B,OAAO,EAAE,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,qBAAqB,EAAE,QAAQ,EAAE,IAAI,EAAE;iBACzE;gBACD,oBAAoB,EAAE,KAAK;aAC5B;YACD,MAAM,EAAE,CAAC,KAAK,EAAE,KAAK,EAAE,EAAE,CAAC,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,iBAAiB,CAAC,KAAK,CAAC,EAAE,CAAC;SAC7E;QACD,KAAK,CAAC,OAAO,CAAC,IAAI,EAAE,IAAI;YACtB,OAAO,OAAO,CAAC,cAAc,CAAC;gBAC5B,OAAO,EAAE,IAAI,CAAC,IAAI;gBAClB,MAAM,EAAE,IAAI,CAAC,MAAM;gBACnB,SAAS,EAAE,WAAW,CAAC,IAAI,CAAC;gBAC5B,OAAO,EAAE,IAAI,CAAC,KAAK,EAAE,OAAO;gBAC5B,MAAM,EAAE,IAAI,CAAC,MAAM;aACpB,CAAC,CAAA;QACJ,CAAC;KACF,CAAC,CAAA;AACJ,CAAC"}
+{"version":3,"file":"profile.js","sourceRoot":"","sources":["../../../src/tools/profile.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,UAAU,EAAE,MAAM,wBAAwB,CAAA;AAEnD,OAAO,EAAE,iBAAiB,EAAsB,MAAM,eAAe,CAAA;AACrE,OAAO,EAAE,qBAAqB,EAAE,MAAM,4BAA4B,CAAA;AAClE,OAAO,EAAE,WAAW,EAAE,MAAM,aAAa,CAAA;AAEzC;;;;GAIG;AACH,MAAM,UAAU,iBAAiB,CAAC,OAA2B;IAC3D,OAAO,UAAU,CAAC;QAChB,IAAI,EAAE,cAAc;QACpB,WAAW,EAAE;YACX,4GAA4G;YAC5G,8UAA8U;YAC9U,4fAA4f;SAC7f,CAAC,IAAI,CAAC,IAAI,CAAC;QACZ,UAAU,EAAE;YACV,IAAI,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE,WAAW,EAAE,2GAA2G,EAAE;YAClK,MAAM,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,WAAW,EAAE,4GAA4G,EAAE;YACrJ,cAAc,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,QAAQ,EAAE,MAAM,EAAE,MAAM,EAAE,aAAa,EAAE,YAAY,EAAE,YAAY,EAAE,WAAW,EAAE,eAAe,EAAE,QAAQ,CAAC,EAAE,WAAW,EAAE,0FAA0F,EAAE;SACjQ;QACD,MAAM,EAAE;YACN,MAAM,EAAE,qBAAqB;YAC7B,MAAM,EAAE,CAAC,KAAK,EAAE,KAAK,EAAE,EAAE,CAAC,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,iBAAiB,CAAC,KAAiC,CAAC,EAAE,CAAC;SACzG;QACD,KAAK,CAAC,OAAO,CAAC,IAAI,EAAE,IAAI;YACtB,OAAO,OAAO,CAAC,cAAc,CAAC;gBAC5B,OAAO,EAAE,IAAI,CAAC,IAAI;gBAClB,MAAM,EAAE,IAAI,CAAC,MAAM;gBACnB,GAAG,CAAC,IAAI,CAAC,cAAc,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,cAAc,EAAE,IAAI,CAAC,cAAc,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;gBACrF,SAAS,EAAE,WAAW,CAAC,IAAI,CAAC;gBAC5B,OAAO,EAAE,IAAI,CAAC,KAAK,EAAE,OAAO;gBAC5B,MAAM,EAAE,IAAI,CAAC,MAAM;aACpB,CAAC,CAAA;QACJ,CAAC;KACF,CAAC,CAAA;AACJ,CAAC"}

package/lib/types/tools/report.d.ts

@@ -0,0 +1,14 @@
+/**
+ * The `data_report` model tool: read persisted profile/clean-diff reports
+ * back from the `data_quality` storage domain, by exact `reportKey` or by
+ * `kind`. Deterministic read-only consumption — no model arithmetic.
+ * @module dsh-data-quality/tools/report
+ */
+import type { DataQualityService } from '../service.js';
+/**
+ * Build the `data_report` tool definition against a mounted service.
+ * @param service - the mounted ctx.dataQuality implementation.
+ * @returns the tool definition to register.
+ */
+export declare function defineReportTool(service: DataQualityService): import("@deepseek-ai/dsh-tools").ToolDefinition;
+//# sourceMappingURL=report.d.ts.map

package/lib/types/tools/report.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"report.d.ts","sourceRoot":"","sources":["../../../src/tools/report.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAMH,OAAO,KAAK,EAAE,kBAAkB,EAAE,MAAM,eAAe,CAAA;AAsDvD;;;;GAIG;AACH,wBAAgB,gBAAgB,CAAC,OAAO,EAAE,kBAAkB,mDAkE3D"}

package/lib/types/tools/report.js

@@ -0,0 +1,112 @@
+/**
+ * The `data_report` model tool: read persisted profile/clean-diff reports
+ * back from the `data_quality` storage domain, by exact `reportKey` or by
+ * `kind`. Deterministic read-only consumption — no model arithmetic.
+ * @module dsh-data-quality/tools/report
+ */
+import { defineTool } from '@deepseek-ai/dsh-tools';
+import { renderCleanHtml, renderProfileHtml } from "../report-html.js";
+/** The report kinds a caller may query. */
+const REPORT_KINDS = ['profile', 'clean', 'clean-diff', 'verify', 'citations'];
+/** Render one stored report as a self-contained HTML document (profile/clean only). */
+function renderRecordHtml(record) {
+    if (record.kind === 'profile') {
+        return renderProfileHtml(record.report);
+    }
+    if (record.kind === 'clean' || record.kind === 'clean-diff') {
+        return renderCleanHtml(record.report, record.dataset);
+    }
+    throw new Error(`data_report html format does not support kind "${record.kind}" (profile/clean only)`);
+}
+/** Project a stored report into the canonical value (the stored report is already lossless JSON). */
+function toView(record) {
+    return { key: record.key, kind: record.kind, at: record.at, dataset: record.dataset, report: record.report };
+}
+/** Human-readable report summary for the tool's Native render. */
+function renderReportText(value) {
+    const lines = [];
+    if (value.key !== undefined) {
+        lines.push(value.records.length === 0 ? `No report for ${value.key}` : `Report ${value.key}`);
+    }
+    else {
+        lines.push(`Reports of kind ${value.kind ?? ''} (${value.records.length})`);
+    }
+    for (const record of value.records) {
+        lines.push(`- ${record.key} [${record.kind}] ${record.dataset} @ ${new Date(record.at).toISOString()}`);
+    }
+    return lines.join('\n');
+}
+/**
+ * Build the `data_report` tool definition against a mounted service.
+ * @param service - the mounted ctx.dataQuality implementation.
+ * @returns the tool definition to register.
+ */
+export function defineReportTool(service) {
+    return defineTool({
+        name: 'data_report',
+        description: [
+            'Read persisted data-quality reports back from the data_quality storage domain (deterministic, read-only).',
+            'Pass key (the exact reportKey a prior run returned) to fetch one report, or kind to list every persisted report of that kind, ordered chronologically. Exactly one of key/kind.',
+            'Returns the report envelope(s): kind, dataset, timestamp, and the full stored report (profile/clean/clean-diff/verify/citations). Missing keys and unknown kinds fail loudly.',
+        ].join('\n'),
+        parameters: {
+            key: { type: 'string', description: 'Exact storage reportKey (e.g. 20260819000000000-profile-1a2b3c4d); fetches that one report.' },
+            kind: { type: 'string', enum: [...REPORT_KINDS], description: 'Report kind to list (profile/clean/clean-diff/verify/citations).' },
+            format: { type: 'string', enum: ['json', 'html'], description: 'Output format. json (default) returns the report envelope(s); html renders one report as a self-contained offline HTML document (requires key; profile/clean only).' },
+        },
+        output: {
+            schema: {
+                type: 'object',
+                properties: {
+                    key: { type: 'string' },
+                    kind: { type: 'string', enum: [...REPORT_KINDS] },
+                    records: {
+                        type: 'array',
+                        items: {
+                            type: 'object',
+                            properties: {
+                                key: { type: 'string', required: true },
+                                kind: { type: 'string', enum: [...REPORT_KINDS], required: true },
+                                at: { type: 'number', required: true },
+                                dataset: { type: 'string', required: true },
+                                report: { type: 'json', required: true },
+                            },
+                            additionalProperties: false,
+                        },
+                        required: true,
+                    },
+                    html: { type: 'string', description: 'Self-contained offline HTML (present only when format: html).' },
+                },
+                additionalProperties: false,
+            },
+            render: (_args, value) => {
+                const view = value;
+                if (view.html !== undefined)
+                    return [{ type: 'text', text: view.html }];
+                return [{ type: 'text', text: renderReportText(view) }];
+            },
+        },
+        async execute(args, _exec) {
+            const hasKey = args.key !== undefined;
+            const hasKind = args.kind !== undefined;
+            if (hasKey === hasKind) {
+                throw new Error('data_report needs exactly one of key/kind');
+            }
+            const format = (args.format ?? 'json');
+            if (hasKey) {
+                const record = await service.getReport(args.key);
+                return {
+                    key: args.key,
+                    records: [toView(record)],
+                    ...(format === 'html' ? { html: renderRecordHtml(record) } : {}),
+                };
+            }
+            if (format === 'html') {
+                throw new Error('data_report html format requires key (exactly one report)');
+            }
+            const records = await service.listReports(args.kind);
+            return { kind: args.kind, records: records.map(toView) };
+        },
+    });
+}
+//# sourceMappingURL=report.js.map