# RELAI

## Docs

- [benchmark](https://cli-docs.relai.ai/cli/benchmark.md): Reference for relai benchmark commands.
- [evaluator](https://cli-docs.relai.ai/cli/evaluator.md): Reference for relai evaluator commands.
- [init](https://cli-docs.relai.ai/cli/init.md): Reference for relai init.
- [learning-env](https://cli-docs.relai.ai/cli/learning-env.md): Reference for relai learning-env commands.
- [optimize](https://cli-docs.relai.ai/cli/optimize.md): Reference for relai optimize.
- [All commands](https://cli-docs.relai.ai/cli/overview.md): The complete public relai surface. Read-only commands are safe to run freely; others write files, upload metadata, remove records, or change code.
- [setup](https://cli-docs.relai.ai/cli/setup.md): Configure the CLI itself, update it, clone the sample, install plugins, and generate shell completions.
- [simulate](https://cli-docs.relai.ai/cli/simulate.md): Reference for relai simulate.
- [tag & memory](https://cli-docs.relai.ai/cli/tag-and-memory.md): Group learning environments into reusable suites, and store durable Markdown guidance RELAI considers during generation and optimization.
- [Compatibility](https://cli-docs.relai.ai/compatibility.md): RELAI is in beta and aims to be framework-agnostic. Here's what your project needs and where support is deepest today.
- [How it works](https://cli-docs.relai.ai/concepts.md): A learning loop that closes itself: failures in, improvements out. The objects that do the work, how simulation and optimization fit together, and the budget math.
- [Verifiable Continual Learning for AI Agents](https://cli-docs.relai.ai/index.md): RELAI turns agent failures and feedback into repeatable learning environments, simulates your agent against them, and optimizes the agent from the results.
- [Install & set up](https://cli-docs.relai.ai/installation.md): Install the CLI, confirm prerequisites, and configure credentials — three commands and you're ready to initialize a project.
- [Try the sample agent](https://cli-docs.relai.ai/onboarding.md): No agent of your own yet? relai onboarding clones a disposable finance-agent sample so you can walk the full loop before pointing RELAI at your own code.
- [Coding-agent plugins](https://cli-docs.relai.ai/plugins.md): RELAI ships plugins that let your coding agent drive the loop for you. Install a local marketplace, then add the RELAI plugin for Codex or Claude Code.
- [Quickstart](https://cli-docs.relai.ai/quickstart.md): The shortest path through the loop on a real project: initialize, create one learning environment, simulate, optimize.

## OpenAPI Specs

- [openapi](https://cli-docs.relai.ai/api-reference/openapi.json)