Agent-readable news context

agentnews.md

Context boards for working AI agents.

Most AI answers get worse when they start from a blank chat. agentnews gives agents a current priority map: what matters now, what is uncertain, which sources support it, and what to search next. It is not a news article site and not a conclusion engine.

Humans can read the HTML pages. Agents can read the same boards as markdown.

About AgentNews explains the problem: AI search can find links, but agents still need a shared, falsifiable now before they answer.

Supported domains

Finance / Macro

Equities, rates, FX, oil, geopolitics, and market plumbing.

• Human HTML: /finance
• Agent markdown: https://agentnews.md/finance.md
• Archive: /finance/archive

Finance / Macro (Korea)

KOSPI/KOSDAQ, the won (USD/KRW), Bank of Korea & rates, Korean macro, and global macro through a Korea lens.

• Human HTML: /finance-ko
• Agent markdown: https://agentnews.md/finance-ko.md
• Archive: /finance-ko/archive

How to read a board

1. Read the domain board before answering a current-domain question.
2. Treat it as a priority map, not a final answer.
3. Preserve uncertainty and evidence labels.
4. For market-facing answers, compare it with current reactions and primary sources.
5. Use follow-up queries and watch threads as search starting points.

Direct access

• HTML for people: https://agentnews.md/finance
• Markdown for agents: https://agentnews.md/finance.md
• Content negotiation: request Accept: text/markdown on the HTML route when supported.