The standalone agent execution engine. Point it at any YAML agent file, provide credentials, and it runs — no server, no database, no UI required.
v1.3.8 · Release notes
Define your agent's name, instructions, connectors, and workflow steps in a simple YAML file. No code required.
Add your LLM API key and connector credentials to oe-config.json. The runtime matches them to your agent automatically.
Execute with a single command — or start as an HTTP API server and call from any app. No Node.js, no Docker, no install.
Pass agent parameters at runtime with --param. Connect to any LLM or service with --config.
Add --serve to turn the binary into a persistent HTTP API. Call agents from mobile apps, web services, or any HTTP client.
/health
Liveness check
/run
Run from inline YAML
/run-file
Run from YAML file on disk
Every capability below is available to any agent via tool calling. Declare the connector in your YAML — the runtime handles the rest.
Each guide includes the full YAML, oe-config.json, and copy-paste run commands for Windows, macOS, and Linux.
Write a YAML agent, add your credentials, run it anywhere. No server required.