🔍 Archival Intelligence: A Forensic Rare Book Auditor
This is a submission for the Notion MCP Challenge
What I Built
I built the Rare Book Intelligence MCP Server, a specialized forensic agent that turns a Notion workspace into an expert appraisal lab. In the world of high-value assets, the difference between a lowercase 'j' and a capital 'J' on a 1925 Gatsby dust jacket represents a $150,000 valuation swing.
This MCP server enables AI agents (like Claude) to navigate a relational graph of four distinct Notion databases—Inventory, Master Bibliography, Market Results, and Audit History—to identify forgeries, verify states, and automate the "Chain of Custody" for rare artifacts.
Video Demo
Archival Intelligence: Solving $150k Rare Book Forgeries with Notion & MCP
Show us the code
Find the full source code, test suite, and prompt library here:
GitHub Repository: kenwalger/Notion_MCP_Challenge_2026
How I Used Notion MCP
Most MCP implementations focus on simple data retrieval. I pushed the Model Context Protocol further by creating a Relational Write-Back Loop:
- Forensic Search: The agent queries the Books Catalog for observed data.
- Archival Collation: It retrieves "Ground Truth" standards from a private Master Bibliography.
- Market Intelligence: It analyzes historical sales in the Market Results database.
- Automated Governance: If a discrepancy is found (like the "j" variant in The Great Gatsby, or the 'wabe/wade' variant in *Alice's Adventures in Wonderland'), the agent uses the Notion API to:
- Flip the inventory status to "Flagged".
- Create a permanent, timestamped Audit Log.
- Relational Linking: Automatically link the log entry back to the catalog item, ensuring a verifiable provenance record.
- Automated Export: Generate a museum-read Exhibit Label in Markdown, synthesizing complex forensic findings into a clear, actionable summary for curators.
This unlocks a forensic "expert-in-your-pocket" workflow for any dealer, collector, or insurance adjuster using Notion.
Reflections & Future Scaling
This project proves that the Model Context Protocol can do more than summarize text—it can enforce Technical Governance. While we used Rare Books as our proof-of-concept, the relational logic is built for the entire multi-billion dollar collectibles market.
At Abiqua Archive, we are scaling this logic to help museums and private collectors automate trust and protect the "Iron and Ink" of our shared history.
Ken W. Alger
Top comments (1)
Building this was a deep dive into how we can move AI from being a 'chatter' to being a 'governance agent.' The most satisfying part was seeing Notion's relational engine handle the forensic back-links automatically—it really proves that Notion + MCP is the future of 'Expert Systems.'
I'd love to hear from other builders: Are you focusing more on data retrieval or are you also exploring these kinds of 'Write-Back' loops for your submissions?