Add 'what the content must cover' brief text area; screenshots stay optional

pipeline/tutorial.py

@@ -63,6 +63,8 @@ Rules:
   limitations where relevant.
 - Output valid JSON only. No markdown, no comments in the actual output.
 
+{brief_block}
+
 {keyword_block}
 
 Transcript (each line is "[mm:ss] text"):
@@ -72,6 +74,21 @@ Transcript (each line is "[mm:ss] text"):
 """
 
 
+def _brief_block(brief: str | None) -> str:
+    """Render the user's 'what the content must cover' brief into the prompt."""
+    brief = (brief or "").strip()
+    if not brief:
+        return ("Content brief: none provided — cover the video's key steps faithfully "
+                "based on the transcript.")
+    return (
+        "Content brief — REQUIRED COVERAGE. The tutorial MUST address every point below. "
+        "Ground each point in the transcript where possible; if the transcript doesn't "
+        "cover a required point, still cover it accurately and concisely and don't invent "
+        "specifics that contradict the video. Organise the steps so these points are "
+        "clearly covered:\n" + brief
+    )
+
+
 def _keyword_block(keywords: dict | None) -> str:
     """Render the SEO/keyword placement guidance injected into the prompt.
 
@@ -185,11 +202,12 @@ def count_keyword(tutorial: dict, keyword: str) -> int:
 
 
 def generate_tutorial(transcript: str, hf_token: str, model: str = DEFAULT_LLM,
-                      keywords: dict | None = None) -> dict:
+                      keywords: dict | None = None, brief: str | None = None) -> dict:
     """Call the chat model and return a normalized tutorial dict.
 
     Returned keys: ``title, slug, meta_description, answer, intro, steps, faqs``.
     ``keywords`` is ``{"primary": str, "secondary": [str, ...]}`` for SEO placement.
+    ``brief`` is the user's free-text 'what the content must cover' requirement.
     """
     if not hf_token:
         raise ValueError("An HF token is required for the tutorial LLM (billed to your key).")
@@ -198,6 +216,7 @@ def generate_tutorial(transcript: str, hf_token: str, model: str = DEFAULT_LLM,
     if len(transcript) > MAX_TRANSCRIPT_CHARS:
         truncated += "\n[... transcript truncated for length ...]"
     prompt = (_INSTRUCTIONS
+              .replace("{brief_block}", _brief_block(brief))
               .replace("{keyword_block}", _keyword_block(keywords))
               .replace("{transcript}", truncated))