Workspace Indexer

Workspace Indexer

Not this skill: Wiki linking, catalog.json, backlinks, and [[slug]] management → use repo-wiki. Not this skill: Single-file text extraction → use to-md.

Overview

This skill provides two core capabilities:

1. Binary → Markdown conversion — Batch-convert PDF, PPTX, DOCX, and XLSX via MarkItDown (see to-md skill policy)
2. Index generation — Build a workspace file inventory and keyword map in index.md

Path conventions

All internal paths mentioned in this skill are relative to the directory containing this SKILL.md, and are not the same as the workspace current directory (./).

• Scripts: scripts/...
• Other resources: relative paths inside the skill directory
• python scripts/... style references must be resolved against this skill's absolute Base Directory at runtime
• In the examples below, <skill-base-dir> means the actual deployed absolute path of this skill
• External workspace targets such as the workspace root must be passed explicitly with flags like --root and --out

Script list

Prerequisites

Conversion follows the to-md skill policy. Install MarkItDown only — do not use pymupdf, python-docx, python-pptx, or openpyxl.

# Automatic install
python <skill-base-dir>/scripts/install_deps.py

# Manual install
pip install "markitdown[all]"

Workflow

Task A: Binary documents → Markdown conversion

Script: scripts/convert_binary_docs.py

# Preview target files with dry-run
python <skill-base-dir>/scripts/convert_binary_docs.py --root . --dry-run

# Convert everything and create .md files next to the originals
python <skill-base-dir>/scripts/convert_binary_docs.py --root .

# Convert only selected formats
python <skill-base-dir>/scripts/convert_binary_docs.py --root . --formats pdf docx

# Write output to a separate directory
python <skill-base-dir>/scripts/convert_binary_docs.py --root . --out ./converted

# Overwrite existing converted files
python <skill-base-dir>/scripts/convert_binary_docs.py --root . --overwrite

Conversion engine: All formats use MarkItDown (convert_with_markitdown in convert_binary_docs.py). Output structure (headings, tables, slides) is determined by MarkItDown — see the to-md skill for details.

Task B: Build index.md

Script: scripts/build_index.py

# Default run (create index.md at the root)
python <skill-base-dir>/scripts/build_index.py --root .

# Preview files and keywords with dry-run
python <skill-base-dir>/scripts/build_index.py --root . --dry-run

# Choose a custom output path
python <skill-base-dir>/scripts/build_index.py --root . --out docs/index.md

# Use a custom keyword file (one keyword per line)
python <skill-base-dir>/scripts/build_index.py --root . --keywords keywords.txt

# Ignore specific directories
python <skill-base-dir>/scripts/build_index.py --root . --ignore-dirs tmp out

index.md structure:

# Workspace Index
> Generated at / Root / File count

## File inventory (text)
### 📁 Directory name
- [filename](path) (size in KB)

## Binary document list
| Filename | Path | Size | Converted file |
(Link the converted `.md` when present; otherwise mark it as not converted)

## Keyword map
### Keyword
- [files where it appears]

Keyword extraction behavior:

• Default: auto-extract from #, ##, and ### headings in .md files
• Custom: map files that contain keywords listed in --keywords keywords.txt

Task C: Full workflow (unified runner)

run.py automatically executes conversion and indexing in sequence.

# Full workflow (recommended)
python <skill-base-dir>/scripts/run.py --root .

# Conversion only (skip indexing)
python <skill-base-dir>/scripts/run.py --root . --skip-index

# Indexing only (skip conversion)
python <skill-base-dir>/scripts/run.py --root . --skip-convert

# Preview the full workflow with dry-run
python <skill-base-dir>/scripts/run.py --root . --dry-run

Task D: Check conversion status

# Show unconverted files in the console
python <skill-base-dir>/scripts/check_status.py --root .

# Show the full status including converted files
python <skill-base-dir>/scripts/check_status.py --root . --show-converted

# Save the status report as Markdown
python <skill-base-dir>/scripts/check_status.py --root . --export conversion_status.md

Notes

• .git, __pycache__, node_modules, .github, and venv directories are ignored by default
• Existing converted files are not regenerated unless --overwrite is set
• Scanned PDFs (image-only) require markitdown[all] for OCR support (see to-md skill)
• For very large workspaces, use --max-keyword-files to cap oversized keyword maps