Orloj CLI
Orloj declares agents, tools and policies in YAML and runs multi-agent systems with queues, approvals and tracing.
- 126 ★ on the repository
- Last commit
- Apache-2.0
Extended access
Why this level
- The platform runs agents with access to tools and secrets.
- The install script is downloaded from the network and executed locally.
Install
- Works in
Manual install
brew tap OrlojHQ/tap && brew install orlojctlThe first CLI install command from the README, and an `install.sh` script also exists.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The platform runs agents with access to tools and secrets.
- The install script is downloaded from the network and executed locally.
This is third-party code. Review the repository files before installing.
What it does
Orloj treats agents as infrastructure services. Agents, tools, models, memory, schedules, webhooks, policies and secrets are YAML resources, controllers drive the system to the declared state, and workers take tasks and carry handoffs between agents with retries and a dead-letter queue. Traces, logs and metrics are visible. The `orlojctl` CLI runs everything: `orlojctl init demo` creates a project, `orlojctl apply -f demo/ --run` launches it and secrets are added with `orlojctl create secret`. Tools can be HTTP, gRPC, WASM, CLI or MCP, and external MCP servers attach through a `McpServer` resource. For Cursor there is an `orloj-generator` skill that writes manifests.
Who it is for. Platform engineering teams that run multi-agent systems as services and want approvals, policies and observability.
Good fit when
- Several agents must hand work to each other with retries and control.
- Human approvals and tool access policies are needed.
- Tasks must run on a schedule or webhook with traces to inspect.
- External MCP servers have to be attached to an agent system as resources.
Not a fit when
- You need a single agent in the terminal without infrastructure.
- You need a stable 1.0 API.
- You are not ready to describe a system with YAML manifests.
Example request
Create a demo project with orlojctl init, add the model key as a secret and run a system of two agents.Limitations
The project is in active development, and before 1.0 the API and resource schemas may change. The platform installs as a set of server components, `orlojctl init demo` is enough for a local look and a full run needs a model provider key. Keys are stored as secret resources, so restrict access to them. The `orloj-generator` skill sits in `.cursor/skills` and is copied by hand for other agents.
How to disable. Uninstall `orlojctl` (`brew uninstall orlojctl`) and stop the deployed components. Resources and secrets in the cluster are removed with `orlojctl` commands.
FAQ
How do I install the CLI?
`brew tap OrlojHQ/tap` and `brew install orlojctl`, or the `install.sh` script from the repository.
Where do I start?
With `orlojctl init demo`, then a secret with the model key and `orlojctl apply -f demo/ --run`.
Does it support MCP?
Yes. A controller discovers tools on attached servers itself and hands them to agents.
What does observability show?
Traces, logs, metrics, the message graph and the approval queue.
Is there a skill for an agent?
Yes, `orloj-generator` in `.cursor/skills` generates resources and manifest schemas for Cursor.
Discussion
No questions or reviews yet.
Related
CLI toolsANOLISA is a system stack for AI agents: output compression, memory, workspace snapshots, command checks and token accounting
A background daemon loops over GitHub or Forgejo issues, drafts specs, reviews and fixes PRs until they are ready for human merge.
A self-hosted gateway between AI apps and models, MCP tools and A2A agents: policy, budgets, routing and an audit trail
A Rust CLI and TUI: agents in containers, parallel tabs, TOML workflows from issue to PR and a scheduled agent squad
Allox CLI creates an isolated agent environment, runs shell, browser and Jupyter, saves checkpoints and restores work.
The official Coralogix CLI with 19 skills: an agent reads production logs, traces and metrics and triages incidents
README in short
The description presents Orloj as a full stack for agent systems: resources, controllers, workers, policies and observability. Component tables and a quick start through `orlojctl` follow.