# Give to Claude: Create the four MVP agents

You are Claude Code running on the Mac/PC where Multica CLI and the Multica daemon are installed.

Your job is to create the four MVP Multica agents for the YDY-style starter pod.

## Safety

- Do not print tokens, auth headers, secrets, or environment values.
- Do not create extra agents beyond the four listed unless the human explicitly asks.
- Do not publish, deploy, spend money, or modify production systems.
- If a Multica command or flag fails, inspect `multica <command> --help` and adapt.
- If login/setup is required, stop and tell the human the exact interactive step needed.

## Step 1 — Check runtime and workspace

Run discovery first:

```bash
claude --version
multica version
multica auth status || true
multica workspace list --output json
multica daemon status
multica runtime list --output json
```

Expected:

- Claude Code is available.
- Multica CLI responds.
- Multica workspace is selected or selectable.
- Daemon is running.
- A Claude Code runtime is visible.

If no runtime exists, ask the human to finish Multica/daemon/runtime setup before continuing.

## Step 2 — Create exactly these agents

Create these four agents, all connected to the same Claude Code runtime:

1. Chief of Staff
2. Researcher
3. SEO Consultant
4. Blog Writer

Use concise descriptions:

- Chief of Staff — coordinates requests, creates tickets, reviews handbacks, returns final decisions.
- Researcher — gathers source material, competitor context, evidence, and buyer language.
- SEO Consultant — audits site/search visibility and recommends prioritized SEO/site actions.
- Blog Writer — writes useful articles, outlines, and content drafts from evidence and briefs.

## Step 3 — Save IDs

Save returned IDs somewhere local, for example:

```bash
mkdir -p /tmp/multica-mvp
cat > /tmp/multica-mvp/agent-ids.env <<'EOF'
CHIEF_OF_STAFF_AGENT_ID="<id>"
RESEARCHER_AGENT_ID="<id>"
SEO_AGENT_ID="<id>"
BLOG_WRITER_AGENT_ID="<id>"
EOF
```

Do not invent IDs. Use the actual Multica output.

## Done when

- The four agents exist in the selected Multica workspace.
- All four use the Claude Code runtime.
- Their IDs are saved locally or clearly reported to the human.
- No extra agents were created.