wrong recognination when paste a block of markdown format text

Nobody has claimed this yet.

### Search first

- [x] I searched and no similar issues were found

### What Happened?

I use this skill to generate logseq markdown format notes:

```markdown
Summarize the previous chat and investigation into a structured Logseq note in English.
Requirements:
Use a single root bullet node with nested structure
The root node must be a single bullet containing a level-1 heading
Example:
- #
category:: #ai-investigation
tags:: #tag1, #tag2
Child nodes should use nested bullets combined with headings for hierarchy (e.g., ##, ###)
Indentation must use TAB characters (not spaces) for nesting
The output must contain exactly one root node
The root node must include:
A concise title
Properties attached directly to the root node (not as separate sibling nodes):
category:: #ai-investigation
tags:: #xxx, #xxx (include relevant tags based on content)
Properties must appear immediately under the root node as its attributes, not as independent nodes
The note should be clear, structured, and easy to scan
The structure should be flexible and adapted to the content type (e.g., bug investigation, question, design discussion, architecture analysis)
Structure guidelines:
Organize content into logical sections based on the context
Prefer commonly useful sections such as (but not limited to):
Summary
Context / Background
Key Findings
Analysis
Decisions
Trade-offs
Evidence
Actions / Next Steps
Do NOT force irrelevant sections
Merge or omit sections when appropriate
Content guidelines:
Use concise bullet points only (no long paragraphs)
Preserve important technical details, reasoning, and conclusions
Capture insights, not just raw facts
Code handling rules (important):
Only wrap content in code blocks if the original chat/investigation contains:
code snippets
commands
logs
Do NOT wrap the entire output in a code block
Keep code snippets in their relevant context
Code reference link rules:
If the investigation refers to a specific code location, file, function, class, or line number, include a Markdown link as a jump point
The link target must use this format: codelink://filepath:linenum
File paths must use forward slashes (/) as separators (not \)
The link text should be a brief human-readable description
The final format should be: [description](codelink://filepath:linenum)
Use the most specific file path and line number available from the original content
File paths must be absolute paths (not relative paths)
If the line number is not available, still include the file path when useful, using the closest available location format
Do not invent file paths or line numbers that are not present in the original content
Diagram rules:
Use PlantUML when a diagram would make the note clearer, especially for:
architecture
component relationships
control flow
algorithm flow
sequence interactions
state transitions
PlantUML diagrams must be included as code blocks so they can be copied directly in Logseq
Use the appropriate PlantUML diagram type based on the content, such as:
sequence diagram
component diagram
activity diagram
state diagram
class diagram
Keep diagrams concise and focused on the key insight
Do not add a diagram when it would be redundant or speculative
Do not invent components, flows, or relationships not supported by the original content
Output constraints:
The entire output MUST be wrapped in a single fenced code block so it can be easily copied into Logseq
Inside the output, code snippets and PlantUML diagrams should still use their own fenced code blocks (nested code blocks are expected)
Do NOT include any explanation before or after the note
Output only the final Logseq-formatted content
```
the old version of logseq can correctly grab the tags and categories of the generated note text
the new version can not

### Reproduce the Bug

use ai to generate a note by the attached skill
paste the skill text into logseq
category and tag setting is not correct

### Expected Behavior

_No response_

### Screenshots

_No response_

### Desktop or Mobile Platform Information

_No response_

### Additional Context

_No response_

### Are you willing to submit a PR? If you know how to fix the bug.

- [ ] I'm willing to submit a PR (Thank you!)

Start by reproducing the paste flow in Logseq with the supplied AI-generated Markdown skill text, then compare how the old and new versions handle the category and tag properties. Done means pasting the note correctly recognizes and applies its category and tags.

Written by the indexing model from the issue text.