hedge-agent: where an agent checks a thesis against what companies said before.

This room belongs to Data Vintage. The agent here works for an investment team, a risk function or a research desk that wants to know whether a pattern held in earlier years: how often a halt preceded a restatement, how long a results date slipped, which venues issued more bulletins. It needs the past as it was published then, with the date it was read, and it needs to show its sources.

1 Who works here

1.1

A research or risk agent that runs studies over many issuers and several years, and must cite every row it used.

1.2

It reads; it never prices, scores or recommends — the study and its conclusions stay with the team that asked.

2 Its method, in order

2.1

Find out which markets reach back how far: list_nodes — the backfill block of each roster entry (history_from, years held, the year in progress).

2.2

Inside the trailing year, pull dated events per issuer with list_events_since and a start date.

2.3

For a record as it stood before a change, ask get_record with a version.

2.4

Keep the receipt of every read (receipt.decision_id); get_receipt at https://mcp.rco-a2a.ai/mcp returns what was read, under which rule, when.

2.5

Rows before the live window are held signed per market and are not yet served on a door; a study that needs them waits for the door or asks the operator.

3 The door it calls

3.1

https://mcp.getissuer.ai/mcp (any market) and each market door listed on the roster; read-only MCP, no account.

4 Paying for a read

$0.01 a read, paid in the call on x402. A study touches thousands of records; the reads are free by default and paid only where the team wants the settled receipt.

5 Machine kit

K.1

/llms.txt · /facts.json · /.well-known/agent-card.json · /.well-known/security.txt · /sitemap.xml · /robots.txt

K.2

The years this room works: data-vintage.ai.

Contact Us

The agents that built this read their own mail: allooloo@hey.com