# Eevo

> Eevo is Upright’s consequential decision model. Send a decision with its options, the consequences you list under each and the stakeholders; in Default mode (recommended for supporting agentic and human judgement) get each option’s expected value per stakeholder, or in Direct mode (for high-volume decisions and custom tasks) each stakeholder’s best option with probabilities in one call. Its SystemOne-compatible API also answers typed questions (choice, score, yes/no) about any state.

## Tutorials

- [Connect your agent to Eevo](https://eevo.io/api/arvo/docs/tutorials/agent.md): Add Eevo’s MCP server to Claude, ChatGPT, Claude Code, Codex or any MCP client, and let your agent ask Eevo.
- [Ask Eevo via curl (Native API)](https://eevo.io/api/arvo/docs/tutorials/first-decision.md): Get an API key, send a decision with its consequences, read each option’s expected value, and try Direct mode.
- [Ask Eevo via curl (SystemOne-compatible API)](https://eevo.io/api/arvo/docs/tutorials/questions.md): Ask typed questions about any state through the SystemOne-compatible API, and act on the probabilities.

## How-to guides

- [Choose Default or Direct mode](https://eevo.io/api/arvo/docs/how-to/choose-mode.md): When to list the consequences yourself, and when to let Eevo weigh the options itself.
- [Write good stakeholders and consequences](https://eevo.io/api/arvo/docs/how-to/stakeholders-and-consequences.md): What to put in the context, which stakeholders to include, and how to list consequences.
- [Report what was done](https://eevo.io/api/arvo/docs/how-to/report-chosen-decisions.md): Tell Eevo which option was carried out, so the console shows whether its advice is followed.
- [Revise a decision and record whether you agree](https://eevo.io/api/arvo/docs/how-to/revise.md): Call again when a result shows something was missed, and record your own verdict.
- [Link decisions to chats, issues and pull requests](https://eevo.io/api/arvo/docs/how-to/metadata.md): Send metadata so your team can find each decision in the log by thread, issue, PR or person.
- [Handle long runs, rate limits and timeouts](https://eevo.io/api/arvo/docs/how-to/retries.md): What a 202 means, which errors to retry, how long to wait, and what is never charged.
- [Configure data controls](https://eevo.io/api/arvo/docs/how-to/retention.md): Choose how long each key’s requests are kept, and whether Eevo may learn from them.

## Reference

- [Authentication](https://eevo.io/api/arvo/docs/reference/authentication.md): API keys, what they pay from, their retention, and the rate limits.
- [API reference](https://eevo.io/api/arvo/docs/reference/api.md): Every endpoint with its fields, generated from the OpenAPI document.
- [Make a decision](https://eevo.io/api/arvo/docs/reference/api-decisions.md): A decision in Default mode (when mode is left out) or Direct mode.
- [Get a decision](https://eevo.io/api/arvo/docs/reference/api-get-decision.md): A decision again, with its chosen decision, revisions and assessments.
- [Share a decision](https://eevo.io/api/arvo/docs/reference/api-sharing.md): Whether everyone in your organization sees a decision, or only you.
- [Report what was done](https://eevo.io/api/arvo/docs/reference/api-chosen.md): The option carried out, for the whole decision.
- [Record whether you agree](https://eevo.io/api/arvo/docs/reference/api-assessments.md): Your verdict on the answer, per stakeholder.
- [Ask typed questions](https://eevo.io/api/arvo/docs/reference/api-systemone.md): The SystemOne-compatible API: typed questions about any state.
- [MCP server](https://eevo.io/api/arvo/docs/reference/mcp.md): Connect Claude and other MCP clients to Eevo, and the tools they get.
- [Agent skills](https://eevo.io/api/arvo/docs/reference/skills.md): Skill files that tell an agent when and how to use Eevo, with a raw URL for each.
- [Models, pricing and limits](https://eevo.io/api/arvo/docs/reference/models-pricing.md): The model, what a request costs, the free decisions, and the request limits.
- [Errors](https://eevo.io/api/arvo/docs/reference/errors.md): Every error has a status, a type your code can branch on, and a message to show.

## Explanation

- [How Eevo decides](https://eevo.io/api/arvo/docs/explanation/how-eevo-decides.md): A model trained to weigh what options lead to for the people they affect, one stakeholder at a time.
- [Probability, confidence and expected value](https://eevo.io/api/arvo/docs/explanation/probability-and-ev.md): What the numbers in an answer mean, and what they do not.
- [Why double counts are discounted](https://eevo.io/api/arvo/docs/explanation/overlap.md): How Eevo finds consequences that count the same effect twice, and what it does about them.

## Agent skills

- [eevo-decision-support](https://eevo.io/api/arvo/docs/skills/eevo-decision-support/SKILL.md): Build the decision with each option’s consequences, run it for every affected stakeholder, revise it when the result shows something was missed, present where the stakeholders agree and conflict, and report what was done.
- [eevo-questions](https://eevo.io/api/arvo/docs/skills/eevo-questions/SKILL.md): Ask narrow typed questions (choice, score, yes/no) and act on the probabilities in code.

## API

- [OpenAPI 3.1](https://eevo.io/api/arvo/openapi.json): every endpoint and field