| name | markdown-outline |
| description | **READING SKILL** — Generate a hierarchical outline index (with line-number ranges) for a long Markdown file, then use it to navigate precisely to sections of interest. USE FOR: reading/understanding large .md documents; answering questions about specific sections without loading the entire file; exploring document structure. DO NOT USE FOR: editing Markdown; non-.md files. |
| argument-hint | Path to the Markdown file to outline and read |
Markdown Outline Index
Generate a structured outline of a long Markdown file with line-number ranges, then use the outline to read only the sections you need. This avoids loading entire large documents into context.
When to Use
- You need to read or answer questions about a long Markdown file (hundreds+ lines)
- You want to understand the structure of a Markdown document before diving in
- You need to find a specific section in a large document efficiently
- User asks about content in a
.md file and you want to navigate precisely
Procedure
Step 1 — Generate the Outline
Run the bundled script on the target file:
python <skill-root>/scripts/get_markdown_outline.py "<target-file>.md"
# Optional: add `--max-level N` to limit heading depth (default 6)
# Optional: add `-o <output-file>` to write the outline to a file instead of stdout
# Example: python ./.agent/skills/markdown-outline/scripts/get_markdown_outline.py "long-document.md" --max-level 3
# Example: python ./.agent/skills/markdown-outline/scripts/get_markdown_outline.py "long-document.md" --max-level 3 -o "long-document.index.md"
- Use
--max-level N (1–6) to limit heading depth if the file is deeply nested. Default is 6.
- The script outputs to stdout by default.
Step 2 — Parse the Outline
The output contains each heading with two key ranges:
| Range | Meaning |
|---|
| Total Span | This heading + all nested sub-headings (the full section) |
| Direct Content | Lines under this heading before the next sub-heading starts |
Example output line:
## 1.1 Topic Name *[Total Span: 10 - 85 | Direct Content: 10 - 15]*
Step 3 — Read Target Sections
Use the line ranges from the outline to read only the sections relevant to the task:
- To read a specific topic and all its sub-sections → use Total Span
- To read just the intro/overview of a section (skipping children) → use Direct Content
- Read multiple non-adjacent sections in parallel using their respective ranges
Tips
- For very large files (1000+ lines), start with
--max-level 2 or --max-level 3 to get a high-level overview, then drill down.
- If the user asks about a topic, match keywords to heading titles in the outline before reading.
- Prefer reading the narrowest range that answers the question.
Script Reference
- Input: Any
.md or .txt file path
- Output: Hierarchical index with
Total Span and Direct Content line ranges
- Options:
--max-level N (1–6), -o [path] (write to file instead of stdout)
