# Creating corpus overview notes

Here's a prompt you can use. Run it once per note or cluster of notes, feeding Claude the content each time.

---

**PROMPT – Replace [X] with your folder path before running it.**

You have access to my Obsidian vault via MCP. Read all notes in the folder [USER MUST SPECIFY FOLDERNAME] and process them into a structured corpus overview.

This overview will be used by an AI content strategist and writer to plan and draft content for ISO27DIY, a B2B SaaS product that helps SMEs implement ISO27001 without hiring consultants.

For each note or cluster of related notes, produce an overview entry in the following format:

---

**Title:** [note title or cluster name] 
**Path:** [filename or folder path — list each note path individually for clusters] 
**Summary:** [2-3 sentences on what this note actually contains — substance, not just topic] 
**Key concepts and terms:** [main concepts, frameworks, or terminology covered] 
**ISO27001 relevance:** [how this connects to ISO27001 implementation, compliance, or cybersecurity practice]
**ISO27DIY relevance:** [how this could support product messaging, content marketing, or user education] 
**Related notes:** [other notes in the vault this connects to, if known]
**Content potential:** [1-2 sentences on what kind of content this could fuel — articles, newsletter topics, LinkedIn posts, forum answers, etc.] 
**Fetch priority:** [High / Medium / Low — how often the content agents are likely to need the full note]

---

Rules:

- Be specific. Vague summaries are useless.
- Do not invent content that isn't in the note. If something is unclear or thin, say so.
- Group closely related notes under one entry but list each path individually.
- Flag any note that seems outdated, incomplete, or too thin to be useful with a [REVIEW] tag after the title.
- Process all notes in the folder before responding. Do not stop after the first note.
---
**How to use it**

Paste the prompt, then paste the raw content of one note or a group of related notes. Run it in batches. Once you have all the entries, compile them into a single Obsidian note called something like `_corpus-overview.md` and upload that to the Project knowledge base.

If your notes are well-tagged or linked in Obsidian, you can also group by tag or folder and process whole clusters at once, which saves a lot of runs.