@eigenpal/cli 0.10.43 → 0.10.45

package/CHANGELOG.md

@@ -1,5 +1,11 @@
 # @eigenpal/cli
 
+## 0.10.45
+
+### Patch Changes
+
+- b44c2fb: The workflow authoring reference now covers explicit OCR and Vision AI parse modes, optional figure-caption models, and structured layout output.
+
 ## 0.10.26
 
 ### Patch Changes

package/dist/cli.js

@@ -19172,7 +19172,8 @@ var init_processor = __esm(() => {
     tenantId: exports_external.string(),
     automationId: exports_external.string().optional(),
     workflowId: exports_external.string(),
-    workerId: exports_external.string().optional()
+    workerId: exports_external.string().optional(),
+    stepExecutionId: exports_external.string().optional()
   });
   ((TraceStatus2) => {
     TraceStatus2["PENDING"] = "pending";
@@ -19409,7 +19410,7 @@ function getSupportedFormatsLabel() {
   const labels = [...new Set(allMimes.map((m) => MIME_TO_LABEL[m] ?? m))];
   return labels.join(", ");
 }
-var OFFICE_MIME_TYPES, ParserCategory, ParserTypeSchema, DocumentMetadataSchema, ParseUsageSchema, CoordinateUnitSchema, PointSchema, BoundingRegionSchema, PositionedTextSchema, TableCellSchema, ExtractedTableSchema, LayoutElementTypeSchema, LayoutElementSchema, PageResultSchema, ParseResultSchema, ParseOptionsSchema, ParserInputSchema, EXTENSION_TO_MIME, MIME_TO_EXTENSION, MIME_TO_LABEL;
+var OFFICE_MIME_TYPES, ParserCategory, ParserTypeSchema, DocumentMetadataSchema, ParseUsageSchema, StructuredDocumentSchema, CoordinateUnitSchema, PointSchema, BoundingRegionSchema, PositionedTextSchema, TableCellSchema, ExtractedTableSchema, LayoutElementTypeSchema, LayoutElementSchema, PageResultSchema, ParseResultSchema, ParseOptionsSchema, ParserInputSchema, EXTENSION_TO_MIME, MIME_TO_EXTENSION, MIME_TO_LABEL;
 var init_parser = __esm(() => {
   init_zod();
   OFFICE_MIME_TYPES = [
@@ -19479,6 +19480,7 @@ var init_parser = __esm(() => {
     pageCount: exports_external.number(),
     processingTimeMs: exports_external.number().optional()
   });
+  StructuredDocumentSchema = exports_external.record(exports_external.string(), exports_external.unknown());
   CoordinateUnitSchema = exports_external.enum(["pixel", "inch", "point", "normalized"]);
   PointSchema = exports_external.object({
     x: exports_external.number(),
@@ -19549,6 +19551,7 @@ var init_parser = __esm(() => {
     parserType: ParserTypeSchema,
     parserVersion: exports_external.string().optional(),
     model: exports_external.string().optional().describe("Model used (for LLM/OCR parsers)"),
+    structured: StructuredDocumentSchema.optional().describe("Canonical structured document with ordered blocks, regions, bounding boxes, tables, figures, and chunks when supported by the parser"),
     rawResponse: exports_external.unknown().optional()
   });
   ParseOptionsSchema = exports_external.object({
@@ -19683,8 +19686,10 @@ var init_document_parser = __esm(() => {
   DocumentParserInputSchema = ResolvedProcessorFileSchema;
   DocumentParserOutputSchema = ParseResultSchema;
   DocumentParserConfigSchema = exports_external.object({
+    parseMode: exports_external.enum(["ocr", "vision"]).optional().describe("Base parser for PDF/image inputs. OCR is the default."),
     ocrModel: exports_external.string().optional().describe("OCR provider ID for PDF/image parsing"),
     llmModel: exports_external.string().optional().describe("LLM provider ID for vision-based parsing"),
+    figureModel: exports_external.string().optional().describe("Vision model used only for the optional figure-description pass"),
     maxConcurrency: exports_external.number().min(1).max(10).default(3).describe("Max concurrent VLM batch requests"),
     pagesPerBatch: exports_external.number().min(1).max(20).default(5).describe("Number of page images per VLM request"),
     pdfRenderScale: exports_external.number().min(1).max(4).default(1).optional().describe("Scale factor used when rendering PDF pages for VLM parsing"),
@@ -65118,12 +65123,15 @@ var AiParseConfigSchema, AiParseOutputSchema, AiExtractConfigSchema, AiExtractOu
 var init_step_configs = __esm(() => {
   init_zod();
   init_common();
+  init_parser();
   init_typed_script();
   init_steps();
   AiParseConfigSchema = exports_external.object({
     input: exports_external.string().describe("Storage reference or template expression for the document"),
+    parseMode: exports_external.enum(["ocr", "vision"]).optional().describe("Base parser for PDF/image inputs. OCR is the default; vision uses the selected LLM. Text and Office files always use native parsing."),
     ocrModel: exports_external.string().optional().describe("OCR provider ID for PDF/image parsing"),
     llmModel: exports_external.string().optional().describe("LLM provider ID for vision-based parsing"),
+    figureModel: exports_external.string().optional().describe("Vision model used only for the optional figure-description pass"),
     maxConcurrency: exports_external.number().min(1).max(10).default(3).optional().describe("Max concurrent VLM batch requests"),
     pagesPerBatch: exports_external.number().min(1).max(20).default(5).optional().describe("Number of page images per VLM request"),
     pdfRenderScale: exports_external.number().min(1).max(4).default(1).optional().describe("Scale factor for rendering PDF pages before VLM parsing. Higher values produce sharper images at larger payload sizes."),
@@ -65135,16 +65143,7 @@ var init_step_configs = __esm(() => {
     describeFigures: exports_external.boolean().optional().describe("Opt-in (default off). After text extraction, detect which pages contain figures with an in-worker layout model, then caption those pages with a vision model and append `<figure>description</figure>` to their text — so image-only pages (property photos, signatures, charts) become findable by text-based steps like ai.split. Note: the layout scan runs over all pages, and the caption step and its vision calls are billed. Skipped for plaintext."),
     figureInstructions: exports_external.string().optional().describe('Custom instruction for the figure-description pass, e.g. "Describe each figure; label a handwritten signature as `<figure>signature</figure>` and a stamp as `<figure>stamp</figure>`; for property photos note the room or exterior shown." Applied only when describeFigures runs.')
   });
-  AiParseOutputSchema = exports_external.object({
-    text: exports_external.string().describe("Extracted text content (combined from all pages)"),
-    pages: exports_external.array(exports_external.object({
-      pageIndex: exports_external.number().describe("0-based page index"),
-      text: exports_external.string().describe("Extracted text for this page"),
-      pageName: exports_external.string().optional().describe("Page/sheet name (e.g., Excel sheet name)"),
-      confidence: exports_external.number().optional().describe("Overall page confidence")
-    })).optional().describe("Per-page content"),
-    metadata: exports_external.record(exports_external.string(), exports_external.unknown()).optional().describe("Document metadata")
-  });
+  AiParseOutputSchema = ParseResultSchema.omit({ rawResponse: true });
   AiExtractConfigSchema = exports_external.object({
     input: exports_external.string().describe("Text content or template expression"),
     schema: exports_external.object({
@@ -184657,7 +184656,7 @@ import { fileURLToPath as fileURLToPath9 } from "url";
 // package.json
 var package_default = {
   name: "@eigenpal/cli",
-  version: "0.10.43",
+  version: "0.10.45",
   description: "Eigenpal CLI — author and run document workflows from your terminal. Agent-ready.",
   keywords: [
     "eigenpal",

package/dist/skill/reference/step-types.md

@@ -365,8 +365,10 @@ Extract text from documents (PDF, DOCX, images) using OCR or vision models
 | Field | Type | Required | Default | Description |
 | --- | --- | --- | --- | --- |
 | `input` | string | yes |  | Storage reference or template expression for the document |
+| `parseMode` | `"ocr"` \| `"vision"` | no |  | Base parser for PDF/image inputs. OCR is the default; vision uses the selected LLM. Text and Office files always use native parsing. |
 | `ocrModel` | string | no |  | OCR provider ID for PDF/image parsing |
 | `llmModel` | string | no |  | LLM provider ID for vision-based parsing |
+| `figureModel` | string | no |  | Vision model used only for the optional figure-description pass |
 | `maxConcurrency` | number | no | `3` | Max concurrent VLM batch requests |
 | `pagesPerBatch` | number | no | `5` | Number of page images per VLM request |
 | `pdfRenderScale` | number | no | `1` | Scale factor for rendering PDF pages before VLM parsing. Higher values produce sharper images at larger payload sizes. |
@@ -382,9 +384,14 @@ Extract text from documents (PDF, DOCX, images) using OCR or vision models
 
 | Field | Type | Required | Default | Description |
 | --- | --- | --- | --- | --- |
-| `text` | string | yes |  | Extracted text content (combined from all pages) |
-| `pages` | array<object> | no |  | Per-page content |
-| `metadata` | record<string, unknown> | no |  | Document metadata |
+| `document` | object | yes |  |  |
+| `usage` | object | yes |  |  |
+| `pages` | array<object> | yes |  |  |
+| `text` | string | yes |  | Combined text from all pages |
+| `parserType` | `"plaintext"` \| `"office"` \| `"llm-vision"` \| `"ocr"` | yes |  |  |
+| `parserVersion` | string | no |  |  |
+| `model` | string | no |  | Model used (for LLM/OCR parsers) |
+| `structured` | record<string, unknown> | no |  | Canonical structured document with ordered blocks, regions, bounding boxes, tables, figures, and chunks when supported by the parser |
 
 #### `ai.extract` — Extract Data
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@eigenpal/cli",
-  "version": "0.10.43",
+  "version": "0.10.45",
   "description": "Eigenpal CLI — author and run document workflows from your terminal. Agent-ready.",
   "keywords": [
     "eigenpal",