jelou commands exist, what each product
area does, and how to chain them, so you can ask it in natural language
“deploy this function” or “build a support flow” and it picks the right
command.
How they work
- On your first successful
jelou login, the CLI detects the installed AI editors and installs the skills globally into them. New sessions in those editors already know aboutjelou. - Each skill is written as markdown under an editor’s skills directory. The
CLI writes a canonical copy to
<root>/.agents/skills/<skill>/and then creates symlinks (or copies, on systems that block symlinks) to the directory of each detected agent. - All skills derive their version from the CLI binary’s version: when
you update
jelou, all skills are reported as updated.
“Skill” here means the markdown package that
jelou agent install writes
into the AI editors. Don’t confuse it with the legacy sense of “skill” as
a workflow within a project.jelou agent install
Install (or reinstall) the skills manually. In an interactive terminal it
asks for target agents, scope, install method, and confirmation. In
CI, JSON mode, agent mode, or with stdout redirected, it never asks.
Supported AI editors
The CLI detects and supports these targets:jelou skills
Inspect and manage the versions of the installed skills.
Available skills
The CLI bundles these skills (several include variants and additional reference files):
In addition, the BrainOps bundles are more complete installable packages (with
executables, references, and hooks) for building and testing workflows:
jelou-build-workflows (building workflows), jelou-test-workflow
(QA for a single workflow), and jelou-audit-workflows (read-only audits).