Obsidian Knowledge Factory

Overview

This skill runs a local file-processing pipeline for Obsidian knowledge ingestion. It replaces webhook workflows with Python modules while keeping traceability, dedupe safety, and archive discipline.

When to Use

• Large batches of notes, articles, transcripts, or PDFs must be converted into atomic notes.
• Existing content has noisy formatting and inconsistent structure.
• You need deterministic output schema and failure queues.
• n8n is unavailable or intentionally removed from the workflow.

Inputs and Outputs

• Input folder: <vault_path>/<inbox_name>
• Output folders: category folders in the same vault
• Archive folder: <vault_path>/<archive_name>
• Failed queue: <vault_path>/inbox/Failed
• Output schema per atom:
  • category
  • filename
  • content
  • source
  • tags
  • confidence

Execution Flow

1. Scan inbox for md/txt/pdf.
2. Read source text.
3. Call OpenAI-compatible model for atomization.
4. Parse and validate JSON output.
5. Deduplicate and merge safely.
6. Write notes and archive source file.

Red Flags

• Missing OPENAI_API_KEY or invalid OPENAI_BASE_URL.
• Workflow role prompt exists but API mode is not enabled.
• Model output cannot be parsed into JSON after cleanup.
• Merge result is empty.
• Source file is archived but no note was written.

Verification Checklist

• Run once: python engine/main.py --once
• Check summary includes scanned/succeeded/failed/merged/archived.
• Confirm failed payloads exist in inbox/Failed when parse errors occur.
• Confirm archived source appears in archive folder only after success.