| name | cooagents-setup |
| description | Install and start cooagents from a cloned repo by calling the unified deployment CLI. Use when the user wants setup, install, bootstrap, or first start. |
| user-invocable | true |
| metadata | {"openclaw":{"emoji":"setup","always":false,"requires":{"bins":["git","node","npm","python"]}},"hermes":{"tags":["cooagents","setup","install"]}} |
Role
You are the setup wrapper for cooagents.
Do not manually replay the old long bootstrap procedure.
Use the repo-local deployment CLI as the single source of truth.
Inputs
Collect only the minimum required inputs:
repo_path: local path to the cloned cooagents repo
repo_url: optional, only needed when the repo does not exist yet
admin_username: optional, default admin
admin_password: required only when auth env is missing or must be replaced
workspace_root: optional, default ~/cooagents-workspace
runtime: none, openclaw, hermes, or both
If the repo is missing and repo_url is present, clone it first.
If the repo is missing and repo_url is absent, ask only for the missing repo location or repo URL.
Flow
- Confirm the repo exists:
exec ls {repo_path}/src/app.py {repo_path}/config/settings.yaml
- If it does not exist and
repo_url is available, clone it:
exec git clone {repo_url} {repo_path}
- Run the unified setup command from inside the repo:
exec cd {repo_path} && python scripts/deploy.py setup --admin-username {admin_username} --admin-password '{admin_password}' --workspace-root '{workspace_root}' --runtime {runtime}
Use runtime=none when the user only wants cooagents itself.
Use runtime=openclaw, runtime=hermes, or runtime=both only when the user explicitly wants notifier/runtime integration on the same machine.
- The CLI is responsible for:
- dependency install and web build
- database initialization
- auth env generation and
.env updates, including AGENT_API_TOKEN
- service start/restart
- health and dashboard validation
- optional OpenClaw/Hermes local integration
Do not duplicate those steps in this Skill unless the CLI fails and you are diagnosing.
Success Criteria
Treat setup as successful only when the CLI completes successfully.
The command itself validates:
/health returns status ok
/ returns HTML
Follow-up
If the user later changes notifier/runtime integration, use:
exec cd {repo_path} && python scripts/deploy.py integrate-runtime --runtime {runtime} --restart-service
If the user needs an explicit manual skill sync, use:
exec cd {repo_path} && python scripts/deploy.py sync-skills
Troubleshooting
Read only when needed:
references/troubleshooting.md
references/hermes-integration.md
Notes
- Repo-first is the canonical happy path.
- This Skill is now a thin wrapper over
python scripts/deploy.py setup.
- If the runtime has not installed this Skill yet, the host agent may read this file directly from the cloned repo path and follow it from there.