Zakros AI Corpus Engine

Evidence-led Linear A research

Explore George Sfougaras’s fixed Zakros v1.5 Linear A evidence graph through read-only server tools: exact inscription and form lookups, conditional parser rules, claim guardrails, coverage and optional individual-sign metadata from the existing Zakros Glyph API. The graph contains 43 nodes, 55 edges, 13 parser rules and 11 guardrails across 16 inscription nodes.

What you can ask

List claims forbidden by the guardrails.
Example path ·

MCP connector

Server URL: https://zakros-corpus-engine.geoalekzander.chatgpt.site/api/mcp
Login: None
Protocol: MCP 2026-07-28
Instructions: Zakros audits a fixed v1.5 evidence graph. Linear A is undeciphered. Keep source records, graph relations, cautious functional inference and speculation separate. Missing records mean absent from this export, not globally unattested. Preserve damaged forms, section boundaries, QA-*310-I vs QA-*118 and OLE vs OLIV. Treat recorded source URLs as pointers, not independent verification. Glyph API labels are metadata, not translations.

Tools

Reads data6

Skills

1 in the ChatGPT listing