command-center-mcp

Local MCP server hosting named, parameterized cross-MCP workflows. Phase 1 ships a LinkedIn reporting pack: six read-only prompts that orchestrate the linkedin-campaign-manager-mcp tools and write artifacts to disk.

Architecture

Prompts-only. The server holds no credentials and never calls external APIs. Each prompt returns a user-message template that tells Claude which downstream MCP tools to call and where to write artifacts. See docs/superpowers/specs/2026-05-09-command-center-mcp-and-li-reporting-pack-design.md.

Install

Option A — from npm (recommended for end users)

npm install -g command-center-mcp

Option B — from source (for development)

git clone <repo-url>
cd command-center-mcp
npm install
npm run build

Register with Claude Code or Claude Desktop

Claude Code

claude mcp add command-center-mcp -s user -- command-center-mcp

Or add manually to ~/.claude.json under mcpServers:

"command-center-mcp": {
  "command": "command-center-mcp"
}

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "command-center-mcp": {
      "command": "command-center-mcp"
    }
  }
}

Required upstream MCPs

The workflows orchestrate other MCPs — install whichever you need for the prompts you'll use:

• LinkedIn workflows → linkedin-campaign-manager-mcp
• Marketo cross-MCP workflows (e.g. cc_li_marketo_lead_reconcile) → a Marketo MCP exposing marketo_get_lead_by_email and marketo_get_program_members

Claude.ai web compatibility

command-center-mcp is currently a stdio MCP — it works with Claude Code and Claude Desktop only. Sharing via Claude.ai Custom Connectors requires hosting it as a remote (Streamable HTTP) MCP, which is on the roadmap but not yet implemented.

Workflows

Run cc_list to see the catalog. Phase 1 + Phase C #1 ship:

• cc_li_monday_brief — last-7d vs prior-7d brief with alerts
• cc_li_month_end_pack — full-month rollup with MoM CPL
• cc_li_creative_dump — per-creative dump with fatigue indicator
• cc_li_targeting_audit — wide CSV of targeting facets per active campaign
• cc_li_lgf_responses_export — LGF responses since a cutoff (PII)
• cc_li_demographics_rollup — spend-weighted demographic rollup across campaigns
• cc_li_marketo_lead_reconcile — cross-MCP: reconcile LGF responses against Marketo (missing / no program / no SFDC / suspended)

Artifacts

Every run writes to ~/command-center/artifacts/YYYY-MM-DD/<workflow>/<HHMMSS>/. Always present: report.md, params.json, meta.json. Workflow-specific CSVs as documented per workflow.

Test

npm test

Adding a new workflow

1. Create src/prompts/<domain>/<name>.ts exporting a PromptModule.
2. Add a unit test alongside it asserting the render output contains the right tool names and artifact paths.
3. Append the import to src/prompts/_registry.ts.
4. npm test && npm run build. No core changes needed.