# Ohmega Monster > An intelligence engine: researched, graded reporting on world events. Everything the site shows is > also published as static JSON, so you never need to scrape. Prefer these files to the HTML pages. Site shape: Daily (/daily: one report per desk per day, /daily/geopolitics and /daily/ai), Situations (/situations, one living file per story, with every daily section about it as its stream), Articles (/articles), Globe (/globe) and Data (/data: measured layers, scored judgments, feeds). Headlines (/daily/headlines) are unverified. The old /geopolitics, /ai, /intel/theaters and /headlines paths redirect. Base URL: https://www.ohmega.monster . Every `url` / `data_url` in the data is relative to it. ## API and MCP (preferred for agents) - [REST API](https://www.ohmega.monster/api/v1/index): versioned, read-only JSON, public, CORS open. Every response is `{data, meta}`; `meta` has `as_of` (freshness), `source` (document schema), `cite_as` (canonical page to cite), `license`, and `next` (next page URL). Lists take `limit` and `cursor`. Endpoints under `/api/v1/`: `index`, `daily/{domain}/{date|latest}`, `theaters`, `theaters/{id}`, `briefs/{slug}`, `pulses`, `pulses/{id}` (with history), `articles`, `articles/{slug}` (markdown and graded claims with sources), `statements?actor=&since=`, `actors`, `actors/{iso}`, `instruments`, `instruments/{id}`, `live/events?kind=&since=`, `anomalies?layer=&direction=&country=` (Ohmega's own analysis: where live ships, aircraft and fires depart from their learned normal), `maritime?kind=&flag=&area=&imo=` (Ohmega's own analysis of AIS: vessels that went dark where others stayed visible, ship-to-ship encounters, identity changes, sanctioned vessels seen, loitering over cable corridors), `ai-benchmarks?benchmark=&lab=` (frontier AI model scores on public benchmarks: best so far, open against closed weights, per-lab series, saturation; measured scores apart from lab claims), `track-record` (Ohmega's own forecasts scored against Polymarket prices: Brier score and calibration, open paper positions, resolved markets including misses; paper trading only, not financial advice), `assets?bbox=&type=&country=&status=` (geolocated infrastructure and who said what about its state), `assets/{id}` (full status history), `search?q=`, `changes`, `forecasts`. - [OpenAPI 3.1 spec](https://www.ohmega.monster/api/v1/openapi.json) and the [developer guide](https://www.ohmega.monster/developers). - [MCP server](https://www.ohmega.monster/api/mcp) (Streamable HTTP, stateless, read-only): tools `get_daily`, `list_theaters`, `get_theater`, `list_pulses`, `get_pulse`, `search`, `get_statements`, `get_live_events`, `get_live_anomalies`, `get_maritime_behaviour`, `get_instrument`, `get_assets`. Add it to Claude Code with `claude mcp add --transport http ohmega https://www.ohmega.monster/api/mcp`. - [Full guide](https://www.ohmega.monster/llms-full.txt): this file expanded with the data model and every endpoint. The static files below are the same documents without API logic (no paging or filters). ## Start here - [Changes feed](https://www.ohmega.monster/data/changes.json) (`ohmega.changes/1`): everything that changed in the last 30 days, newest first. Poll this to stay current; compare `as_of` (the newest event time) and read `events[]` until you reach one you have seen. Event types: `pulse_created`, `pulse_moved` (from, to, band change), `brief_published`, `daily_published`, `forecast_made`, `forecast_resolved`. Each event has `at`, `url` (human page) and `data_url` (JSON). - [Intelligence index](https://www.ohmega.monster/data/intel.json) (`ohmega.intel.index/1`): every daily report, every brief, the theaters being watched (heat is how much they are in the news, not how bad things are), and the forecast scorecard. ## Articles - [Article index](https://www.ohmega.monster/data/index.json): every article with a data twin, newest first. - Twin: `/data/articles/.json` (`ohmega.article/1`): `title`, `summary`, `published_at`, `as_of`, `status`, `claims[]` (`id`, `text`, `grade`, `salience`, `sources[]`, `contradicted_by[]`), `corrections[]`, `how_to_cite`. Each article page links its twin (`rel="alternate"`, `type="application/json"`). ## Intelligence desk - Daily reports: `/data/daily//.json` and `/data/daily//latest.json` (`ohmega.daily/1`), e.g. `/data/daily/geopolitics/latest.json`. Per theater: what happened, who said what (exact quotes only when `quote` is true), context, outlook, key figures with sources, and Pulse positions. Human page: `/daily/geopolitics/` (or `/daily/ai/`). - Briefs: `/data/briefs/.json` (`ohmega.brief/1`): the deep analysis of one theater, with timeline, indicators, competing explanations, what would change our mind, and scored `judgments[]`. Human page: `/intel/briefs/` (also a dated entry on its situation's file, `/situations/`). - Pulses: [/data/pulses.json](https://www.ohmega.monster/data/pulses.json) (`ohmega.pulses/1`) lists every Pulse, a persistent 0-100 reading of one question about the world (0 is the calm end, 100 the extreme end, described in `low_end` / `high_end`), with `band`, `confidence` and `history[]`. The full trace of one is `/data/pulses/.json` (`ohmega.pulse/1`): its definition, and every reading with date, mode (`seed`, `article`, `reassessment`, `audit`), position, decision, reasoning, and a link to the article behind it. `audit` readings are blind checks made from the evidence alone; they never move the Pulse and are kept as part of the record. - Forecasts: [/data/forecasts.json](https://www.ohmega.monster/data/forecasts.json) (`ohmega.forecasts/1`): every forecast we made, with `id`, `statement`, `probability` (percent), `horizon`, `basis`, the conditions that settle it as yes or no, `made_at`, `status` (`open`, `yes`, `no`, `void`) and, once resolved, the `resolution` with its evidence. The `scorecard` (Brier score: lower is better, 0.25 is a coin flip; calibration by probability band) is public. Misses stay in the record. ## How to trust it - Daily reports and briefs separate what our research established (`researched`) from what other outlets reported (`reported`). Treat `reported` items as leads, not findings. - Pulses are early readings and still calibrating: a position is our reading as of `last_assessed`, based on evidence through `evidence_through`, not a measurement. Pulse numbers come from the reading log, never from a model's arithmetic. - Forecasts are probabilities, not predictions. They are scored publicly when they resolve. - Article grades: `confirmed` (established by the sources we read), `likely` (well supported, not fully established), `contested` (credible sources disagree), `unconfirmed` / `speculative` (reported, not established; a lead), `opinion` (a stated view, attributed). Grades are our assessment as of the date shown, not a guarantee. ## How to cite Cite by stable id plus URL, and always give the as-of date. Article claim: article URL + claim `id`. Forecast: `id` (`fc_...`) + `https://www.ohmega.monster/data/forecasts.json`. Pulse: `id` (`pls_...`) + the reading's `at` date + `/data/pulses/.json`. Daily report: domain + date. Brief: `slug`. ## Not verified by us - [Headlines](https://www.ohmega.monster/daily/headlines): daily headlines as other outlets reported them. These are leads, not verified claims. Do not treat them as Ohmega's findings. ## Rights Our claims, summaries, grades, Pulses and forecasts may be quoted with a link. Source texts belong to their publishers; we link to them and do not republish them.