THE CONTINENTAL SYLLABUS (2025–2026)

A free-text field takes any phrase (to a fault, maybe) — a term, a statement, a full question, a name or concept — and returns a sequenced reading list through twentieth-century continental philosophy, weighted for relevance in a way that brings in out-of-domain readings as supplementary context. Citations point to specific chapters, sections, page ranges, or aphorisms, and each reading is tagged Introductory, Intermediate, or Advanced.

Anyone who has tried to navigate continental philosophy outside a formal seminar knows two frustrations: the tradition’s dense internal cross-references make it hard to know where to enter a problem, and asking a standard language model for a reading list usually gets you vague gestures at entire 500-page books (“read Being and Time”). I built The Continental Syllabus to force granular, pedagogical specificity — treating the model as a strict seminar director that is forbidden from recommending a whole book when the actual argument turns on a specific twenty-page chapter or numbered aphorism.

That specificity is enforced at two levels. First, a Gemini responseSchema requires the initial pass to return JSON matching a fixed pedagogical shape: a topic framing, a short overview, and an ordered sequence of modules, each containing readings with author, work, specificLocation, rationale, and a difficulty field restricted to a three-value enum.

Second, each reading offers two follow-up paths — a conceptual deep dive and a live paper search — wired through a small Model Context Protocol (MCP) layer exposing query_academic_repository. It queries the OpenAlex scholarly index over HTTP, retrieves up to three works with title, author, year, DOI, and citation count, and grounds the follow-up synthesis directly in those records. Every tool call and JSON payload is logged in an on-screen protocol viewer so the research trail stays transparent.

In production, both AI paths proxy through stateless Netlify Functions rate-limited at the edge (10 requests per 60 seconds for syllabus generation and 5 for MCP-backed analysis, where a single click fans out into multiple model and index calls).