Cournot Intelligence

A Probability For Anything

An agent skill that returns one number for an event, the claim it was computed against, and the sources behind it.

First release covers crypto events  ·  three free calls per account

Install

One command.

Cournot Intelligence is a skill. Install it once in the agent runtime you already use, and your agent can call it when you ask.

01

Install

Run the command once, in the agent runtime you already use.

02

Ask

/cournot followed by your question, or "use Cournot to look this up". The skill runs only on an explicit request — it is never triggered on its own, and one instruction produces at most one paid call.

03

Read the answer

Your agent restates the claim, the probability and the sources. It adds nothing that the API did not return.

Install
$ npx cournot-skills add Cournot-AI/cournot-skills/skills/cournot
Free tier
3 calls per account
After that
paid per call
Trigger
explicit request only
What one call returns

One number,
and what it rests on.

Every answer carries the claim it was computed against and the sources that informed it. If the request cannot be resolved to a specific claim, the skill stops and asks rather than guessing.

Agent sessionrules v1.0
> /cournot will BTC close above $150,000 by year end
Probability 34%
Claim answered
BTC/USD daily close > $150,000
2026-12-31 23:59 UTC
Basis
· Market-implied distribution — primary input
· Cross-source consensus — reference check
· Positioning signal — tail adjustment
Rules v1.0 · An assessment of probability. Not investment advice.

The claim is restated in full

You ask in your own words. The answer states the proposition it was actually computed against, including threshold and deadline.

Sources are listed with their role

Each input appears with the part it played. Weights and the intermediate figures behind each input are not returned.

Credentials stay outside the model

Payment is handled by a local client. The skill never requests, reads or relays wallet credentials, and they do not enter the model's context.

How the number is produced

01

Collect

The probabilities implied for the same event across several independent sources.

02

Combine

Cournot's valuation model weights those inputs into a single estimate. The model is not published.

03

Return

One probability, the claim it answers, and the sources that informed it.

Get started

Three free calls per account.

Install the skill and ask a question. Calls beyond the free tier are paid per call.