deploy

---

name: deploy description: Deploy a merged PR or commit to an environment. TRIGGER when: chained by workflow-deploy or another skill; user explicitly invokes /deploy with a PR number, commit SHA, or ref. SKIP: interactive guided deployments (use workflow-deploy). allowed-tools: Bash AskUserQuestion argument-hint: "[PR-number or ref] [environment]" model: haiku disable-model-invocation: true

Step 0 — Load configuration

Check if /tmp/ops-suite-session/config.json exists:

• If yes, read it (pre-parsed by session-start hook).
• If no, read ${XDG_CONFIG_HOME:-$HOME/.config}/ops-suite/config.yaml (preferred) or the plugin's config.yaml (legacy), parse it, and write to /tmp/ops-suite-session/config.json for other skills to reuse. If neither exists, tell the user to run /ops-suite:configure (or copy config.example.yaml to ~/.config/ops-suite/config.yaml and fill in their values). Stop here.

Extract:

• deploy.ci_provider — determines which adapter to load
• deploy.image_tag_source — how to find the image tag (run-id, commit-sha, tag)
• deploy.migration_tool — whether migrations need to run
• environments — available environments

Also read the reference at references/commands.md (in this skill's directory) for the step-by-step process.

Step 1 — Load adapter

Read the adapter file at adapters/{deploy.ci_provider}.md (in this skill's directory). If the adapter does not exist, tell the user that the CI provider {deploy.ci_provider} is not yet supported and stop.

Step 2 — Identify the change to deploy

If $ARGUMENTS contains a PR number:

• Use the adapter to check the PR state (must be merged)
• Extract the merge commit SHA

If $ARGUMENTS contains a commit SHA or branch ref:

• Verify it exists and is reachable from the main branch

If nothing is provided, ask the user for a PR number or ref.

Step 3 — Determine target environment

If $ARGUMENTS contains an environment name, use it. Otherwise, list available environments from config and ask the user.

IMPORTANT: If the CI workflow auto-deploys to dev on merge (check the build run for a deploy-dev job), inform the user that dev was already deployed automatically. Then:

1. Ask the user to confirm dev is healthy before proceeding to prod.
2. Do NOT deploy to prod until the user explicitly confirms dev is working.
3. If the user asks to deploy directly to prod without verifying dev, warn them and ask for confirmation.

Never chain merge → prod deploy without explicit approval for each environment.

Step 4 — Find the build artifact

Using the adapter's commands and deploy.image_tag_source:

• run-id: Find the CI run associated with the merge commit, extract the run ID
• commit-sha: Use the merge commit SHA directly as the image tag
• tag: Find the latest git tag on the merge commit

Verify the build completed successfully before proceeding.

Step 5 — Check current state

Use the adapter to show:

1. What is currently deployed in the target environment
2. What will change (diff between current and new)

Step 6 — Confirm deployment

ALWAYS ask the user for explicit confirmation before deploying.

Display:

Ready to deploy:
  Change:      PR #{pr_number} — {pr_title}
  Image tag:   {image_tag}
  Target:      {environment}
  Current tag: {current_tag}

Proceed? (yes/no)

Step 7 — Deploy

Use the adapter's deploy command to trigger the deployment. Monitor the deployment progress using the adapter's status commands.

Step 8 — Post-deploy verification

After deployment completes, run the following read-only checks automatically:

1. Use ops-suite:service-status with arguments: {service} {env_name}. Use session state from /tmp/ops-suite-session/ — do not re-ask for environment.

   • If unhealthy → report immediately

2. Use ops-suite:service-logs with arguments: {service} {env_name}. Check for errors in last 5 minutes.

   • If errors found → report but continue

Step 9 — Report and suggest next steps

Deployment Summary:
  PR:          #{pr_number} — {pr_title}
  Environment: {environment}
  Image tag:   {image_tag}
  Status:      {success/failed}
  Duration:    {time}

Post-deploy checks:
  Service health: {ok/degraded}
  Error check:    {clean/errors found}

If deploy.migration_tool is not "none", add:

Next steps:
  → Run `/ops-suite:db-migrate {env_name}` to check and apply pending migrations.