From db07f7453070bf67b21dc895630f4b7d1a3793a4 Mon Sep 17 00:00:00 2001
From: Haitao Pan <manbuzhe2009@qq.com>
Date: Tue, 2 Jun 2026 15:59:31 +0800
Subject: [PATCH] Tighten illustrated PDF workflow acceptance

---
 .../content-pdf-with-images/SKILL.md          | 39 ++++++++++++++++++-
 1 file changed, 38 insertions(+), 1 deletion(-)

diff --git a/skills/workflows/content-pdf-with-images/SKILL.md b/skills/workflows/content-pdf-with-images/SKILL.md
index f707005..1034906 100644
--- a/skills/workflows/content-pdf-with-images/SKILL.md
+++ b/skills/workflows/content-pdf-with-images/SKILL.md
@@ -7,6 +7,9 @@ description: "图文 PDF 工作流：将主题或文章拆成章节，为每章
 
 This is an orchestration skill. It does not replace planning, image, or PDF skills. It forces a staged handoff so the final artifact is a real PDF with real image inputs, not a placeholder file.
 
+Hard rule: image prompts are not images. A PDF that contains only chapter text
+and image prompts is incomplete and must not be reported as delivered.
+
 ## Output Contract
 
 Final deliverables must stay inside the current XWorkmate/OpenClaw task artifact scope:
@@ -48,10 +51,17 @@ Do not claim completion after this phase.
 
 Use the image-series skill path for one standalone PNG per chapter.
 
+If an image-generation tool is available, call it for each chapter before PDF
+assembly. If no image-generation tool is available, stop after `article.md` and
+`prompts/image-prompts.md`, then report that the PDF cannot be completed because
+the required image artifacts are missing.
+
 Required checks before continuing:
 
 - PNG count equals chapter count.
 - Every `assets/images/*.png` is a real non-empty PNG.
+- Every chapter image is a generated or user-provided bitmap image, not a text
+  placeholder, blank rectangle, SVG prompt card, or decorative filler.
 - `assets/images/manifest.md` exists and references only relative paths inside the artifact scope.
 - `prompts/image-prompts.md` and `series.config.json` exist.
 
@@ -70,6 +80,29 @@ The PDF must contain:
 
 Use a PDF library such as reportlab when available. The file must be large enough to plausibly contain text and images; a tiny one-page placeholder PDF is a failure.
 
+Do not use a Markdown-to-PDF command by itself unless the Markdown contains real
+image links and the generated PDF is verified to contain embedded image objects.
+
+After creating the PDF, verify at least one embedded image object exists and the
+image count is at least the chapter count. With `pypdf`, use an equivalent check:
+
+```python
+from pypdf import PdfReader
+
+reader = PdfReader("exports/final.pdf")
+image_count = 0
+for page in reader.pages:
+    resources = page.get("/Resources", {})
+    xobjects = resources.get("/XObject", {})
+    if hasattr(xobjects, "get_object"):
+        xobjects = xobjects.get_object()
+    for obj in xobjects.values():
+        target = obj.get_object()
+        if target.get("/Subtype") == "/Image":
+            image_count += 1
+assert image_count >= EXPECTED_CHAPTER_COUNT, image_count
+```
+
 ## Phase 5: Delivery Check
 
 Before final response, run equivalent checks:
@@ -81,4 +114,8 @@ pdfinfo exports/final.pdf
 pdftotext exports/final.pdf - | head -40
 ```
 
-Only report success if `exports/final.pdf` exists and the expected source files are present.
+Also run an embedded-image check equivalent to the `pypdf` snippet in Phase 4.
+
+Only report success if `exports/final.pdf` exists, the expected source files are
+present, and the PDF contains embedded images. If the PDF has no embedded image
+objects, report it as incomplete and keep the missing-image issue visible.