davidmosiah/timeline-pulse

Timeline Pulse is an open source project that acts as both an MCP server and an interactive website for exploring folklore about timeline shifts, Mandela Effect reports, and myths that grew up around the Large Hadron Collider being shut down for maintenance in mid 2026. The README is upfront that this is a cultural archive, not an attempt to prove anything supernatural. Official facts and community beliefs are kept on separate shelves, and every claim in the corpus is labeled with how solid its source is. The project ships as an npm package called timeline-pulse that can be added to any MCP-compatible AI agent setup with a short config snippet, no API keys or backend required. It comes with a starter dataset built in: 20 reports, 17 recurring symbols, 6 Mandela Effect catalog items, 8 story threads, and 9 timeline events anchored to real CERN announcements. Once connected, an AI agent gets a set of read only tools for searching reports, pulling up Mandela Effect variants, mapping recurring symbols, comparing belief versus skeptical framings of a topic, and getting corpus statistics. The same dataset also powers a public website with a searchable archive, a timeline view, an interactive map of recurring symbols, a Mandela Effect catalog, and story threads that can be read through a believer, neutral, or skeptic lens. Raw JSON data files are also exposed directly for anyone who wants to build on top of them. Each claim in the dataset carries an evidence grade, ranging from official institutional statements down through firsthand accounts, community reports, interpretive readings, symbolic language, and flagged speculation. The project has automated checks in its test suite that enforce this policy, for example requiring that official claims link to real sources. Setup for development uses standard npm commands to install dependencies, run tests, build the server, and sync data into the website. The code is released under the MIT license, while original written summaries and metadata are covered by a separate content policy described in the project's documentation.