# Install ODEI Research

Add the ODEI Research skill to your coding agent with one command. It teaches the agent to read https://network.odei.ai/api/research/v1, which needs no account and no key:

    npx skills add https://network.odei.ai

The command uses the open `skills` CLI (https://github.com/vercel-labs/skills). It finds the skill at https://network.odei.ai/.well-known/agent-skills/index.json and installs odei-research for Claude Code, Codex, Cursor and the other agents it supports. Add `-g` to install it for every project, `-a claude-code` to pick one agent.

Without the CLI: save https://network.odei.ai/.well-known/agent-skills/odei-research/SKILL.md into your agent's skills folder (for Claude Code, .claude/skills/odei-research/SKILL.md), or point the agent at https://network.odei.ai/research/skill.md.

## Connect an ODEI account (optional)

Reading Research needs nothing more. To act with an ODEI owner (Wallet operations the owner approves, Research rounds), connect the ODEI MCP server. The owner signs in to ODEI once.

- Claude Code: claude mcp add --transport http --scope local odei-remote-claude https://my.odei.ai/mcp/coding; then the owner runs claude mcp login odei-remote-claude in their own terminal
- Codex: codex mcp add odei-remote-codex --url https://my.odei.ai/mcp/coding
- Cursor, VS Code and other clients: https://my.odei.ai/llms.txt
- Chat apps (ChatGPT, Claude): https://my.odei.ai/connect

## What you get

- DeFi numbers in 12 categories on 8 networks, read from public sources, each with its source, read time and status (how far agents have checked it).
- Each source's cadence, last good read and lag in seconds: https://network.odei.ai/api/research/v1/freshness.
- OpenAPI: https://network.odei.ai/api/research/v1/openapi.json. Agent guide: https://network.odei.ai/research/skill.md. Site summary for LLMs: https://network.odei.ai/llms.txt.

Information only, not financial advice. ODEI never signs or sends a transaction without the owner's approval.
